29 Mei 2023

Untuk tugas pekan ini, anda diminta untuk mengumpulkan minimal 3 paper dengan tema/keyword (pilih salah satu): deep learning, convolutional neural network (CNN), atau recurrent neural network (RNN).

Anda bisa cari paper di alamat web doaj.org. Maksimal tahun paper yg anda kumpulkan adalah 7 tahun terakhir (terbitan 2016-2023), tidak boleh sebelum 2016. Setiap orang harus mengumpulkan 3 paper yg berbeda dg mahasiswa yg lain. Tidak boleh 1 paper dikumpulkan oleh 2 orang atau lebih. Anda bisa lihat paper2 yg sudah dikumpulkan rekan anda dari komentar2 sebelum anda posting.

Kemudian tuliskan ABSTRACT dari paper yang anda kumpulkan di kolom komentar. Sebelum tulisan abstract, tuliskan: Penulis, tahun, judul paper, penerbit, halaman (jika ada). 

Kerjakan tugas ini paling lambat tgl 3 Juni jam 23.59.


Contoh cara menjawab tugas di kolom komentar blog ini:

Paper: He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770-778).

Abstract: ..... (anda kutip dari paper tsb).

[Kelas, Nama, NPM]




44 komentar:

Anonim mengatakan...

Nama : Bayu Aristianto
NPM : 51419303
Kelas : 4IA07
----------------------------------------------------------------------------
Howard, Andrew & Zhu, Menglong & Chen, Bo & Kalenichenko, Dmitry & Wang, Weijun & Weyand, Tobias & Andreetto, Marco & Adam, Hartwig. (2017). MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications.

Kami menghadirkan kelas model efisien yang disebut MobileNets untuk aplikasi mobile dan embedded vision. MobileNets didasarkan pada arsitektur ramping yang menggunakan konvolusi yang dapat dipisahkan secara mendalam untuk membangun jaringan saraf dalam yang ringan. Kami memperkenalkan dua hyper-parameter global sederhana yang memperdagangkan latensi dan akurasi secara efisien. Hyper-parameter ini memungkinkan pembuat model untuk memilih model dengan ukuran yang tepat untuk aplikasi mereka berdasarkan batasan masalah. Kami menghadirkan eksperimen ekstensif pada pengorbanan sumber daya dan akurasi dan menunjukkan kinerja yang kuat dibandingkan dengan model populer lainnya pada klasifikasi ImageNet. Kami kemudian mendemonstrasikan keefektifan MobileNets di berbagai aplikasi dan kasus penggunaan termasuk deteksi objek, klasifikasi finegrain, atribut wajah, dan geo-lokalisasi berskala besar.
----------------------------------------------------------------------------
Rawat, Waseem & Wang, Zenghui. (2017). Deep Convolutional Neural Networks for Image Classification: A Comprehensive Review. Neural Computation. 29. 1-98. 10.1162/NECO_a_00990.

Convolutional neural networks (CNNs) telah diterapkan untuk tugas-tugas visual sejak akhir 1980-an. Namun, terlepas dari beberapa aplikasi yang tersebar, mereka tidak aktif hingga pertengahan 2000-an ketika perkembangan dalam daya komputasi dan munculnya data berlabel dalam jumlah besar, dilengkapi dengan algoritme yang ditingkatkan, berkontribusi pada kemajuan mereka dan membawa mereka ke garis depan jaringan saraf. renaisans yang telah melihat perkembangan pesat sejak 2012. Dalam ulasan ini, yang berfokus pada penerapan CNN untuk tugas klasifikasi gambar, kami membahas perkembangannya, dari pendahulunya hingga sistem pembelajaran mendalam mutakhir. Sepanjang jalan, kami menganalisis (1) keberhasilan awal mereka, (2) peran mereka dalam kebangkitan pembelajaran mendalam, (3) karya simbolik terpilih yang telah berkontribusi pada popularitas mereka baru-baru ini, dan (4) beberapa upaya perbaikan dengan meninjau kontribusi dan tantangan. lebih dari 300 publikasi. Kami juga memperkenalkan beberapa tren mereka saat ini dan tantangan yang tersisa.
----------------------------------------------------------------------------
Hakim, Heba & Fadhil, Ali. (2021). Survey: Convolution Neural networks in Object Detection. Journal of Physics: Conference Series. 1804. 012095. 10.1088/1742-6596/1804/1/012095.

Dalam beberapa tahun terakhir, jaringan saraf dalam diamati sebagai yang paling berpengaruh di antara semua inovasi di bidang visi komputer, menghasilkan kinerja yang luar biasa pada klasifikasi gambar. Convolutional neural networks (CNN) dianggap sebagai alat yang menarik untuk mempelajari penglihatan biologis karena kategori sistem penglihatan buatan ini menunjukkan kemampuan pengenalan visual yang mirip dengan pengamat manusia. Dengan meningkatkan kinerja pengenalan model ini, tampaknya mereka menjadi lebih efektif dalam prediksi. Tolok ukur terbaru menunjukkan bahwa CNN yang dalam adalah pendekatan yang sangat baik untuk pengenalan dan deteksi objek. Dalam makalah ini, kami berfokus pada blok bangunan inti dari arsitektur jaringan saraf konvolusi. Berbagai metode deteksi objek yang memanfaatkan jaringan saraf konvolusi dibahas dan dibandingkan. Di sisi lain, ada ringkasan sederhana dari arsitektur CNN umum.

Anonim mengatakan...

Nama : Farhan Septian Hidayat
NPM : 52419267
Kelas : 4IA04
Christin, S., Hervet, É., & Lecomte, N. (2019). Applications for deep learning in ecology. Methods in Ecology and Evolution, 10(10), 1632-1644.
Abstract
A lot of hype has recently been generated around deep learning, a novel group of artificial intelligence approaches able to break accuracy records in pattern rec‐ ognition. Over the course of just a few years, deep learning has revolutionized several research fields such as bioinformatics and medicine with its flexibility and ability to process large and complex datasets. As ecological datasets are becoming larger and more complex, we believe these methods can be useful to ecologists as well. In this paper, we review existing implementations and show that deep learning has been used successfully to identify species, classify animal behaviour and esti‐mate biodiversity in large datasets like camera‐trap images, audio recordings and videos. We demonstrate that deep learning can be beneficial to most ecological disciplines, including applied contexts, such as management and conservation. We also identify common questions about how and when to use deep learning, such as what are the steps required to create a deep learning network, which tools are available to help, and what are the requirements in terms of data and computer power. We provide guidelines, recommendations and useful resources, including a reference flowchart to help ecologists get started with deep learning. We argue that at a time when automatic monitoring of populations and eco‐ systems generates a vast amount of data that cannot be effectively processed by humans anymore, deep learning could become a powerful reference tool for ecologists.

Sakib, S., Ahmed, N., Kabir, A. J., & Ahmed, H. (2019). An overview of convolutional neural network: its architecture and applications.
Abstract
With the increase of the Artificial Neural Network (ANN), machine learning has taken a forceful twist in recent times [1]. One of the most spectacular kinds of ANN design is the Convolutional Neural Network (CNN). The Convolutional Neural Network (CNN) is a technology that mixes artificial neural networks and up to date deep learning strategies. In deep learning, Convolutional Neural Network is at the center of spectacular advances. This artificial neural network has been applied to several image recognition tasks for decades [2] and attracted the eye of the researchers of the many countries in recent years as the CNN has shown promising performances in several computer vision and machine learning tasks. This paper describes the underlying architecture and various applications of Convolutional Neural Network.
Liu, P., Qiu, X., & Huang, X. (2016). Recurrent neural network for text classification with multi-task learning. arXiv preprint arXiv:1605.05101.
Abstract
Neural network based methods have obtained great progress on a variety of natural language processing tasks. However, in most previous works, the models are learned based on single-task supervised objectives, which often suffer from insufficient training data. In this paper, we use the multitask learning framework to jointly learn across multiple related tasks. Based on recurrent neural network, we propose three different mechanisms of sharing information to model text with task-specific and shared layers. The entire network is trained jointly on all these tasks. Experiments on four benchmark text classification tasks show that our proposed models can improve the performance of a task with the help of other related tasks.

Anonim mengatakan...

Name: Sis Adrian Luthfi
NPM: 56419085
Kelas: 4IA04

----------------------------------------------------------------------------
https://doi.org/10.1002/ail2.31

Benedict W. J. Irwin, Thomas M. Whitehead, Scott Rowland, Samar Y. Mahmoud, Gareth J. Conduit, Matthew D. Segall. (2021). Deep Imputation on large-scale drug discovery data
Keywords: AI in drug discovery, deep learning, drug discovery, imputation

Abstrak :
Prediksi yang akurat terhadap sifat biologis senyawa kimia membantu dalam penemuan obat, mengatasi biaya tinggi, dan meningkatkan tingkat keberhasilan dalam penelitian farmasi. Namun, metode kecerdasan buatan (AI) menghadapi tantangan akibat keterbatasan data dan kebisingan dalam hasil eksperimen biologis. Dalam makalah ini, kami menunjukkan bagaimana deep learning dalam imputasi data mengungguli model QSAR yang umum digunakan. Dengan menggunakan dataset sebanding dengan repositori perusahaan farmasi, kami menemukan perbaikan signifikan dalam tiga area aplikasi praktis. Metode deep learning mencapai koefisien determinasi median R2 yang lebih tinggi (0,69, 0,36, dan 0,43) dibandingkan metode QSAR random forest (0,28, 0,19, dan 0,23). Kami juga menunjukkan bahwa estimasi yang kuat terhadap ketidakpastian nilai prediksi berkorelasi dengan akurasi, memberikan kepercayaan dalam pengambilan keputusan berdasarkan nilai-nilai yang diimputasi.

---------------------------------------------------------------------------
https://doi.org/10.1002/ail2.33

Alden Dima, Sarah Lukens, Melinda Hodkiewicz, Thurston Sexton, Michael P. Brundage. (2021). Adapting Natural Language Processing for Technical Text
Keywords: Domain Adaption, Maintenance Records, Natural Language Processing, Technical Data, Technical Language Processing

Abstrak :
Pemrosesan bahasa alami (NLP) belum siap mengatasi berbagai masalah dunia nyata karena ketergantungan pada korpus besar, paradigma pembelajaran dangkal, dan kebutuhan sumber daya komputasi yang besar. Makalah ini mengusulkan pemrosesan bahasa teknis (TLP) yang mengintegrasikan rekayasa ke dalam NLP untuk mengekstraksi informasi dari bahasa para ahli dalam domain teknis. TLP melihat NLP sebagai sistem sosio-teknis dan mengatasi masalah data dalam rekayasa. Kami menjelaskan perbedaan pendekatan TLP terhadap makna dan generalisasi, serta manfaat inklusi pengetahuan dari data tak terstruktur. Pendekatan TLP ini diilustrasikan dalam studi kasus pemeliharaan industri.

----------------------------------------------------------------------------
https://doi.org/10.1002/ail2.34

Ejay Nsugbe, Ibrahim Sanusi. (2021). Towards an Affordable Magnetomyography Instrumentation and Low Model Complexity Approach for Labour Imminency Prediction using a Novel Multiresolution Analysis.
Keywords: Cybernetics, Kernel Methods, Logistic Regression, Obstetrics, Optimisation, Preterm

Abstrak :
Prediksi awitan persalinan penting dalam pengaturan klinis. Magnetomiografi menunjukkan potensi dalam memprediksi persalinan, namun terbatas oleh konsumsi sumber daya yang tinggi. Dalam penelitian ini, digunakan lima saluran elektroda dan algoritma dekomposisi sinyal baru untuk mengklasifikasikan persalinan dalam rentang waktu 0-48 jam dan >48 jam. Hasil penelitian menunjukkan bahwa dengan representasi parsimonius, biaya model prediksi berbasis magnetomiografi dapat lebih terjangkau dan memiliki tingkat interpretabilitas yang baik. Metode dekomposisi baru meningkatkan hasil penelitian sekitar 20% dan menggunakan kelompok fitur yang lebih sedikit untuk regresi logistik dan mesin vektor pendukung.

Naufal M Mufhli mengatakan...

Nama : Naufal Muhammad Mufhli
Kelas : 4IA04
NPM : 54419726
Paper 1 : Musab Coskun,Ozal Yildirim, Aysegul Ucar, Yakup Demir (2017), An Overview Of Popular Deep Learning Methods
Abstract: This paper offers an overview of essential concepts in deep learning, one of the state of the art approaches in machine learning, in terms of its history and current applications as a brief introduction to the subject. Deep learning has shown great successes in many domains such as handwriting recognition, image recognition, object detection etc. We revisited the concepts and mechanisms of typical deep learning algorithms such as Convolutional Neural Networks, Recurrent Neural Networks, Restricted Boltzmann Machine, and Autoencoders. We provided an intuition to deep learning that does not rely heavily on its deep math or theoretical constructs.
Key words: Deep Learning, Convolutional Neural Networks, Recurrent Neural Networks, Autoencoders, Restricted Boltzmann Machines

Paper 2: John Gamboa(2017), Deep Learning for Time-Series Analysis
Abstract : In many real-world application, e.g., speech recognition or sleep stage classification, data are captured over the course of time, constituting a Time-Series. Time-Series often contain temporal depen dencies that cause two otherwise identical points of time to belong to different classes or predict different behavior. This characteristic gener ally increases the difficulty of analysing them. Existing techniques often depended on hand-crafted features that were expensive to create and re quired expert knowledge of the field. With the advent of Deep Learning new models of unsupervised learning of features for Time-series analysis and forecast have been developed. Such new developments are the topic of this paper: a review of the main Deep Learning techniques is pre sented, and some applications on Time-Series analysis are summaried. The results make it clear that Deep Learning has a lot to contribute to the field.
Keywords: Artificial Neural Networks, Deep Learning, Time-Series

Paper3: Lucas von Chamier, Romain F. Laine, Johanna Jukkala, Christoph Spahn, Daniel Krentzel, Elias Nehme, Martina Lerche, Sara Hernandez-Perez, Pieta K. Mattila, Eleni Karinou, Seamus Holden, Ahmet Can Solak, Alexander Krull, Tim-Oliver Buchholz, Martin L. Jones, Loic A.Royer, Christophe Leterrier, Yoav Scechtman, Florian Jug, Mike Heilemann, Guillaume jacquemet, Ricardo Henriques(2021), Democrasting Deep Learning for Microscopy with ZeroCostDL4Mic.
Abstract : eep Learning (DL) methods are powerful analytical tools for microscopy and can outper form conventional image processing pipelines. Despite the enthusiasm and innovations fuelled by DL technology, the need to access powerful and compatible resources to train DL networks leads to an accessibility barrier that novice users often find difficult to overcome. Here, we present ZeroCostDL4Mic, an entry-level platform simplifying DL access by lever aging the free, cloud-based computational resources of Google Colab. ZeroCostDL4Mic allows researchers with no coding expertise to train and apply key DL networks to perform tasks including segmentation (using U-Net and StarDist), object detection (using YOLOv2), denoising (using CARE and Noise2Void), super-resolution microscopy (using Deep-STORM), and image-to-image translation (using Label-free prediction - fnet, pix2pix and CycleGAN). Importantly, we provide suitable quantitative tools for each network to evaluate model performance, allowing model optimisation. We demonstrate the application of the platform to study multiple biological processes

Anonim mengatakan...

Nama : Muhammad Raihan
NPM : 54419354
Kelas : 4IA04
Paper 1
Ni, Y., Mao, J., Fu, Y., Wang, H., Zong, H., & Luo, K. (2023). Damage Detection and Localization of Bridge Deck Pavement Based on Deep Learning. Sensors, 23(11), 5138.
https://doi.org/10.3390/s23115138
Abstract :
Bridge deck pavement damage has a significant effect on the driving safety and long-term durability of bridges. To achieve the damage detection and localization of bridge deck pavement, a three-stage detection method based on the you-only-look-once version 7 (YOLOv7) network and the revised LaneNet was proposed in this study. In stage 1, the Road Damage Dataset 202 (RDD2022) is preprocessed and adopted to train the YOLOv7 model, and five classes of damage were obtained. In stage 2, the LaneNet network was pruned to retain the semantic segmentation part, with the VGG16 network as an encoder to generate lane line binary images.
Paper 2
Wan Z, Wan J, Cheng W, Yu J, Yan Y, Tan H, Wu J. A Wireless Sensor System for Diabetic Retinopathy Grading Using MobileViT-Plus and ResNet-Based Hybrid Deep Learning Framework. Applied Sciences. 2023; 13(11):6569.
https://doi.org/10.3390/app13116569
Abstract :
Traditional fundus image-based diabetic retinopathy (DR) grading depends on the examiner’s experience, requiring manual annotations on the fundus image and also being time-consuming. Wireless sensor networks (WSNs) combined with artificial intelligence (AI) technology can provide automatic decision-making for DR grading application. However, the diagnostic accuracy of the AI model is one of challenges that limited the effectiveness of the WSNs-aided DR grading application. Regarding this issue, we propose a WSN architecture and a parallel deep learning framework (HybridLG) for actualizing automatic DR grading and achieving a fundus image-based deep learning model with superior classification performance, respectively.
Paper 3
Jing, L., Zhong, Q., Li, X., Wang, X., Shen, L., & Cao, Y. (2023). Using Deep Learning to Identify Circulation Patterns of Intense Rainfall in the Beijing–Tianjing–Hebei Region. Atmosphere, 14(6), 930.
https://doi.org/10.3390/atmos14060930
Abstract :
The properties and distributions of precipitation are often determined by specific synoptic patterns. Hence, the objective identification of corresponding impact patterns is an important field of research for improving rain forecasting. However, the identification of the weather patterns producing intense rainfall is much more challenging. Since they are violent and local, impact patterns tend to be meso- or smaller-scale systems and are often incompletely presented or only presented in limited regions. In this paper, a deep learning network with a feature cross-fusion module, FConvNeXt, was proposed to address this difficulty and showed great potential. Four major patterns corresponding to intense rainfall in the Beijing–Tianjing–Hebei Region were studied. Statistical testing showed that FConvNeXt performed better than ConvNeXt and ResNet and that the model could identify the weak synoptic forcing type, the subtropical high-pressure type, and the low-vortex pattern with high accuracy.

Silvia Anggraeni S mengatakan...
Komentar ini telah dihapus oleh pengarang.
Silvia Anggraeni S mengatakan...

Nama : Silvia Anggraeni Suganda
NPM : 56419067
Kelas : 4IA07

Theodore Warsavage, Fuyong Xing, Anna E Barón, William J Feser, Erin Hirsch, York E Miller, Stephen Malkoski, Holly J Wolf, David O Wilson, Debashis Ghosh (2020). Quantifying the incremental value of deep learning: Application to lung nodule detection. Published in PLoS ONE. p. e0231468.

Abstrak:
We present a case study for implementing a machine learning algorithm with an incremental value framework in the domain of lung cancer research. Machine learning methods have often been shown to be competitive with prediction models in some domains; however, implementation of these methods is in early development. Often these methods are only directly compared to existing methods; here we present a framework for assessing the value of a machine learning model by assessing the incremental value. We developed a machine learning model to identify and classify lung nodules and assessed the incremental value added to existing risk prediction models. Multiple external datasets were used for validation. We found that our image model, trained on a dataset from The Cancer Imaging Archive (TCIA), improves upon existing models that are restricted to patient characteristics, but it was inconclusive about whether it improves on models that consider nodule features. Another interesting finding is the variable performance on different datasets, suggesting population generalization with machine learning models may be more challenging than is often considered.


Yangling Ma, Yixin Luo (2021). Bone fracture detection through the two-stage system of Crack-Sensitive Convolutional Neural Network. Published in Informatics in Medicine Unlocked (Elsevier). p. 100452.

Abstrak:
Automated fracture detection is an essential part in a computer-aided tele-medicine system. Fractures often occur in human's arbitrary bone due to accidental injuries such as slipping. In fact, many hospitals lack experienced surgeons to diagnose fractures. Therefore, computer-aided diagnosis (CAD) reduces the burden on doctors and identifies fracture. We present a new classification network, Crack-Sensitive Convolutional Neural Network (CrackNet), which is sensitive to fracture lines. In this paper, we propose a new two-stage system to detect fracture. Firstly, we use Faster Region with Convolutional Neutral Network (Faster R-CNN) to detect 20 different types of bone regions in X-ray images, and then we recognize whether each bone region is fractured by using CrackNet. Total of 1052 images are used to test our system, of which 526 are fractured images and the rest are non-fractured images. We assess the performance of our proposed system with X-ray images from Haikou People's Hospital, achieving 90.11% accuracy and 90.14% F-measure. And our system is better than other two-stage systems.


Chongqing Shi, Xiaoli Zhang (2023). Recurrent neural network wind power prediction based on variational modal decomposition improvement. Published in AIP Advances. pp.025027–025027-8.

Abstrak:
In order to avoid the problem that the traditional recurrent neural network (RNN) wind power prediction model cannot take into account both the law of wind power variation and the impact of sudden change factors, this paper proposes an improved cyclic neural network wind power prediction model based on variational modal decomposition (VMD). The VMD algorithm is used to decompose the output power of wind power into different frequency components and analyze the impact of different frequency components on the prediction model. Combined with the feature extraction ability of the neural network, it can reduce the impact of abrupt abnormal data on the prediction results and improve the real-time prediction accuracy of wind power. According to the historical data of an actual wind farm, the results show that the accuracy of the wind power prediction model based on VMD and the recurrent neural network is more than 85%, which is superior to the traditional RNN and the standard long short term memory wind power prediction model.

Asep Hikmat Dermawan mengatakan...

Nama : Asep Hikmat Dermawan
NPM : 51419094
Kelas : 4IA07

Paper 1: Jin-Yeol Kwak, Yong-Joo Chung (2020). Sound Event Detection Using Derivative Features in Deep Neural Networks. Published in Applied Sciences (MDPI AG) (p. 4911).

Abstract: We propose using derivative features for sound event detection based on deep neural networks. As input to the networks, we used log-mel-filterbank and its first and second derivative features for each frame of the audio signal. Two deep neural networks were used to evaluate the effectiveness of these derivative features. Specifically, a convolutional recurrent neural network (CRNN) was constructed by combining a convolutional neural network and a recurrent neural networks (RNN) followed by a feed-forward neural network (FNN) acting as a classification layer. In addition, a mean-teacher model based on an attention CRNN was used. Both models had an average pooling layer at the output so that weakly labeled and unlabeled audio data may be used during model training. Under the various training conditions, depending on the neural network architecture and training set, the use of derivative features resulted in a consistent performance improvement by using the derivative features. Experiments on audio data from the Detection and Classification of Acoustic Scenes and Events 2018 and 2019 challenges indicated that a maximum relative improvement of 16.9% was obtained in terms of the F-score.

Paper 2 : Chunyu Zhao, Haiyang Hou, Qiongying Gu (2022). The Types of Learning Approaches Used by Engineering Students in Three Scenarios: An Adaptation of the R-SPQ-2F to China. Published in Frontiers in Psychology (Frontiers Media S.A.).

Abstract: Deep learning is a type of high-level learning that has received widespread attention in research on higher education; however, learning scenarios as an important variable have been ignored to some extent in past studies. This study aimed to explore the learning state of engineering students in three learning scenarios: theoretical learning, experimental learning, and engineering practice. Samples of engineering university students in China were recruited online and offline; the students filled in the engineering Education-Study Process Questionnaire, which was revised from the R-SPQ-2F. The results of clustering analysis showed four types of learning approaches in the three scenarios: typical deep learning, typical shallow learning, deep-shallow learning, and free learning. Engineering learners in different learning scenarios tended to adopt different learning approaches and showed gender differences. Due to factors such as differences in culture and choice of learning opportunities, the deep and shallow learners demonstrated excellent learning performance, which is in sharp contrast with the “learning failure” exhibited by such students abroad.

Paper 3: Yihao Bai, Weidong Cheng, Weigang Wen, Yang Liu (2022). Fault Diagnosis of Rotating Machinery Based on Evolutionary Convolutional Neural Network. Published in Shock and Vibration (Hindawi Limited) (pp. 45 – 54).

Abstract: This paper proposes a fault diagnosis method for rotating machinery based on evolutionary convolutional neural network (ECNN). With the time-frequency images as the network input, with the help of the global optimization ability of the genetic algorithm, the structure of the convolutional neural network can evolve autonomously, and the adaptive configuration of the structural hyperparameters for the target task is realized. In this paper, the proposed method is verified by the measured signal of the planetary gearbox. The results show that the proposed method is helpful to obtain a convolutional neural network structure with better performance and achieve higher fault diagnosis accuracy.
[4IA07, Asep Hikmat Dermawan, 51419094]

Lia Listriani mengatakan...

Name : Lia Listriani
NPM : 53419403
Kelas : 4IA07

Paper 1 : Jiapeng Yan, Huifang Kong, Zhihong Man. (2022). Recurrent Neural Network-Based Nonlinear Optimization for Braking Control of Electric Vehicles. Published in Energies (MDPI AG) (p. 9486).

Abstract:
In this paper, electro-hydraulic braking (EHB) force allocation for electric vehicles (EVs) is modeled as a constrained nonlinear optimization problem (NOP). Recurrent neural networks (RNNs) are advantageous in many folds for solving NOPs, yet existing RNNs’ convergence usually requires convexity with calculation of second-order partial derivatives. In this paper, a recurrent neural network-based NOP solver (RNN-NOPS) is developed. It is seen that the RNN-NOPS is designed to drive all state variables to asymptotically converge to the feasible region, with loose requirement on the NOP’s first-order partial derivative. In addition, the RNN-NOPS’s equilibria are proved to meet Karush–Kuhn–Tucker (KKT) conditions, and the RNN-NOPS behaves with a strong robustness against the violation of the constraints. The comparative studies are conducted to show RNN-NOPS’s advantages for solving the EHB force allocation problem, it is reported that the overall regenerative energy of RNN-NOPS is 15.39% more than that of the method for comparison under SC03 cycle.

Paper 2 : Nuan Wen, Fang Zhang (2020). Extended Factorization Machines for Sequential Recommendation. Published in IEEE Access (pp. 41342 – 41350)

Abstract:
Users' historical activities are usually contained in real-life sequential recommendation systems to predict their future behaviors. In this situation, traditional Factorization Machines (FMs) approaches may be not suitable. Recently, a new surge of interest aims to use recurrent neural networks(RNN) to encode users' dynamic features with temporal characteristics. However, most of these works fail to reproduce computational simplicity of FMs. In this paper, we propose an architecture of extended-FM for sequential recommendation, which presents temporal feature interactions in an explicit way as traditional FM's formula. Our approach also allows us to accomplish computation of the model in linear time. Furthermore, we merge extended-FM into higher-order interaction framework without significant changes to the deeper models themselves. We conduct comprehensive experiments on two real-world datasets. The results demonstrate that extended-FM outperforms traditional FMs as well as deep learning feature combination models on sequential recommendation tasks.

Paper 3 : Kaixuan Zhang, Qinglong Wang, C. Lee Giles (2021). An Entropy Metric for Regular Grammar Classification and Learning with Recurrent Neural Networks. Published in Entropy (MDPI AG). (p. 127).

Abstract:
Recently, there has been a resurgence of formal language theory in deep learning research. However, most research focused on the more practical problems of attempting to represent symbolic knowledge by machine learning. In contrast, there has been limited research on exploring the fundamental connection between them. To obtain a better understanding of the internal structures of regular grammars and their corresponding complexity, we focus on categorizing regular grammars by using both theoretical analysis and empirical evidence. Specifically, motivated by the concentric ring representation, we relaxed the original order information and introduced an entropy metric for describing the complexity of different regular grammars. Based on the entropy metric, we categorized regular grammars into three disjoint subclasses: the polynomial, exponential and proportional classes. In addition, several classification theorems are provided for different representations of regular grammars. Our analysis was validated by examining the process of learning grammars with multiple recurrent neural networks. Our results show that as expected more complex grammars are generally more difficult to learn.

[4IA07, Lia Listriani, 53419403]

I Putu Cahya Adi Ganesha mengatakan...

I Putu Cahya Adi Ganesha, 52419866, 4IA04

Paper: Zhang, Y., Wei, Y., Jiang, D., Zhang, X., Zuo, W., & Tian, Q. (2023). ControlVideo: Training-free controllable text-to-video generation. In arXiv preprint arXiv:2305.13077.
Abstract: Model difusi yang didorong oleh teks telah membuka kemampuan luar biasa dalam generasi gambar, tetapi generasi video masih tertinggal karena biaya pelatihan yang tinggi untuk pemodelan temporal. Selain beban pelatihan, video yang dihasilkan juga mengalami ketidakkonsistenan penampilan dan kedipan struktural, terutama dalam sintesis video panjang. Untuk mengatasi tantangan ini, peneliti merancang kerangka kerja tanpa pelatihan yang disebut ControlVideo untuk memungkinkan generasi teks-ke-video yang alami dan efisien. ControlVideo, yang diadaptasi dari ControlNet, memanfaatkan konsistensi struktural kasar dari urutan gerak masukan dan memperkenalkan tiga modul untuk meningkatkan generasi video. Pertama, untuk memastikan kohesi penampilan antarbingkai, ControlVideo menambahkan interaksi lintas bingkai penuh dalam modul perhatian diri. Kedua, untuk mengurangi efek kedipan, peneliti memperkenalkan penghalus bingkai berselang yang menggunakan interpolasi bingkai pada bingkai bergantian. Terakhir, untuk menghasilkan video panjang dengan efisien, peneliti menggunakan penyampel hirarkis yang secara terpisah mensintesis setiap klip pendek dengan kohesi holistik. Dengan diperkuat oleh modul-modul ini, ControlVideo mengungguli metode terkini secara kuantitatif dan kualitatif pada berbagai pasangan gerak-teks.

Paper: Borsos, Z., Sharifi, M., Vincent, D., Kharitonov, E., Zeghidour, N., & Tagliasacchi, M. (2023). SoundStorm: Efficient parallel audio generation. In arXiv preprint arXiv:2305.09636.
Abstract: Mempersembahkan SoundStorm, sebuah model untuk generasi audio yang efisien dan non-autoregresif. SoundStorm menerima sebagai input token semantik dari AudioLM, dan mengandalkan perhatian bidireksional dan dekoding paralel berbasis confidence untuk menghasilkan token dari kodek audio neural. Dibandingkan dengan pendekatan generasi autoregresif dari AudioLM, model SoundStorm menghasilkan audio dengan kualitas yang sama dan konsistensi yang lebih tinggi dalam suara dan kondisi akustik, dan dua kali lebih cepat. SoundStorm dapat menghasilkan 30 detik audio dalam 0,5 detik pada TPU-v4. SoundStorm menunjukkan kemampuan model untuk meningkatkan generasi audio dengan urutan yang lebih panjang dengan mensintesis segmen dialog alami berkualitas tinggi.

Paper: Xu, X., Guo, J., Wang, Z., Huang, G., Essa, I., & Shi, H. (2023). Prompt-free diffusion: Taking “text” out of text-to-image diffusion models. In arXiv preprint arXiv:2305.16223.
Abstract: Penelitian teks-ke-gambar (T2I) berkembang pesat dalam setahun terakhir, berkat model difusi yang telah dilatih secara besar-besaran (pre-trained) dan banyak solusi pendekatan personalisasi dan pengeditan yang muncul. Namun, satu hal yang kurang tetap ada: text prompt engineering, dan mencari teks prompt berkualitas tinggi untuk hasil yang dikustomisasi lebih merupakan seni tersendiri. Selain itu, seperti yang sering dikatakan: “sebuah gambar berharga seribu kata” - usaha untuk mendeskripsikan gambar yang diinginkan dengan teks seringkali berakhir menjadi ambigu dan tidak dapat mencakup detail visual yang halus secara menyeluruh, sehingga memerlukan kontrol tambahan dari domain visual. Dalam makalah ini, peneliti mengambil langkah berani ke depan dengan menghilangkan “Teks” dari model T2I difusi yang telah dilatih sebelumnya, untuk mengurangi upaya rekayasa permintaan yang merepotkan bagi pengguna. Kerangka kerja yang diusulkan, Prompt-Free Diffusion, mengandalkan hanya masukan visual untuk menghasilkan gambar baru: ia menerima gambar referensi sebagai “konteks”, pengkondisian struktural gambar opsional, dan noise awal, tanpa teks permintaan sama sekali. Arsitektur inti di balik layar adalah Semantic Context Encoder (SeeCoder), yang menggantikan enkoder teks berbasis CLIP atau LLM yang biasa digunakan.

deviayulestari mengatakan...

Nama : devi ayu lestari
kelas : 4IA04
NPM : 51419702
------------------------------------------------------------------------------------------

Paper 1:
Jha, S., Prashar, D., Long, H. V., & Taniar, D. (2020). Recurrent neural network for detecting. Elsevier Ltd.
Abstrak:
Dalam makalah ini, kami mengusulkan Recurrent Neural Network (RNN) yang efisien untuk mendeteksi malware. RNN adalah klasifikasi jaringan saraf tiruan yang terhubung antar node untuk membentuk grafik terarah bersama dengan urutan temporal. Dalam makalah ini, kami telah melakukan beberapa percobaan dengan menggunakan nilai parameter hiper yang berbeda. Dari percobaan kami yang teliti, kami menemukan bahwa ukuran langkah adalah faktor yang lebih penting daripada ukuran input saat menggunakan RNN untuk klasifikasi malware. Untuk membenarkan pembuktian konsep RNN sebagai pendekatan yang efisien untuk deteksi malware, kami mengukur kinerja RNN dengan tiga vektor fitur berbeda menggunakan parameter hiper. Tiga vektor fitur adalah "vektor fitur enkode panas", "vektor fitur acak", dan "vektor fitur Word2Vec". Kami juga melakukan uji-t berpasangan untuk menguji hasil jika mereka signifikan satu sama lain. Hasil kami menunjukkan bahwa, RNN dengan vektor fitur Word2Vec mencapai nilai Area Under the Curve (AUC) tertinggi dan varians yang baik di antara tiga vektor fitur. Dari analisis empiris, kami menyimpulkan bahwa RNN dengan vektor fitur yang dimiliki oleh arsitektur Skip-gram dari model Word2Vec adalah yang terbaik untuk deteksi malware dengan performa dan stabilitas tinggi.
Keywords: Area under the curve (AUC), Recurrent neural network (RNN), Malware detection, Text classification, Word2Ve
https://doi.org/10.1016/j.cose.2020.102037


Paper 2:
MOGHAR, A., & HAMICHE, M. (2020). Stock Market Prediction Using LSTM Recurrent Neural Network. International Workshop on Statistical Methods and Artificial Intelligence (IWSMAI 2020), 1169-1173.
Abstrak:
Dalam makalah ini, Recurrent Neural Networks (RNN) telah diterapkan untuk mengklasifikasikan denyut normal dan abnormal pada EKG. Tujuan utama dari makalah ini adalah untuk memungkinkan pemisahan otomatis ketukan teratur dan tidak teratur. Database MIT-BIH Arrhythmia sedang digunakan untuk mengklasifikasikan kinerja klasifikasi ketukan. Metodologi yang digunakan dilakukan dengan menggunakan data standar volume besar yaitu data deret waktu EKG sebagai input ke Jaringan Memori Jangka Pendek Panjang. Kami membagi dataset sebagai pelatihan dan pengujian sub-data. Keefektifan, akurasi, dan kemampuan metodologi deteksi aritmia ECG kami ditunjukkan dan perbandingan kuantitatif dengan model RNN yang berbeda juga telah dilakukan.
Keywords: Long Short Term Memory; Recurrent Neural Network ; Arrhythmia; Gated Recurrent unit;
https://doi.org/10.1016/j.procs.2018.05.045

paper 3:
shraddha, s., Pandey, S. K., Pawar, U., & Jangheld, R. R. (2018). Classification of ECG Arrhythmia using Recurrent Neural Networks. International Conference on Computational Intelligence and Data Science (ICCIDS 2018), 1291-1297.
Abstrak:
Tidak pernah mudah untuk berinvestasi dalam satu set aset, pasar keuangan yang tidak normal tidak memungkinkan model sederhana untuk memprediksi nilai aset masa depan dengan akurasi yang lebih tinggi. Pembelajaran mesin, yang membuat komputer melakukan tugas-tugas yang biasanya membutuhkan kecerdasan manusia saat ini menjadi tren dominan dalam penelitian ilmiah. Artikel ini bertujuan untuk membangun model menggunakan Recurrent Neural Networks (RNN) dan khususnya model Long-Short Term Memory (LSTM) untuk memprediksi nilai pasar saham di masa depan. Tujuan utama dari makalah ini adalah untuk melihat presisi mana yang dapat diprediksi oleh algoritma pembelajaran Mesin dan seberapa besar zaman dapat meningkatkan model kami.
Keywords: Recurrent Neural Network Long Short-Term Memory Stock Market forecasting prediction
https://doi.org/10.1016/j.procs.2020.03.049

Twdtri mengatakan...

Nama : Tri Wulandari
Kelas : 4IA16
Npm : 56419398

Recurrent Neural Network (RNN)

Paper 1
Guiyoung Son, Soonil Kwon, and Neungsoo Park (2019). Gender Classification Based on the Non-Lexical Cues of Emergency Calls with Recurrent Neural Networks (RNN). Front. Psychol. 12:742172

https://doi.org/10.3390/sym11040525

Abstrak: Klasifikasi gender otomatis dalam ucapan adalah bidang penelitian yang menantang dengan berbagai aplikasi dalam HCI (interaksi manusia dan komputer). Beberapa dekade penelitian telah menunjukkan hasil yang menjanjikan, tetapi masih ada kebutuhan untuk perbaikan. Sampai sekarang, klasifikasi gender telah dilakukan dengan menggunakan perbedaan dalam karakteristik spektral antara pria dan wanita. Kami berasumsi bahwa ada batas netral antara rentang spektral pria dan wanita. Batas ini menyebabkan kesalahan klasifikasi gender. Untuk mengatasi keterbatasan ini, kami mempelajari tiga fitur ucapan non-lexical (pengisi, tumpang tindih, dan pemanjangan). Dari analisis statistik, kami menemukan bahwa tumpang tindih dan pemanjangan efektif dalam klasifikasi gender. Selanjutnya, kami melakukan klasifikasi gender menggunakan tumpang tindih, pemanjangan, dan fitur akustik dasar, yaitu Koefisien Cepstral Frekuensi Mel (MFCC). Kami mencoba mencapai hasil terbaik dengan menggunakan kombinasi fitur yang berbeda secara bersamaan atau berurutan. Kami menggunakan dua jenis metode pembelajaran mesin, yaitu mesin vektor pendukung (SVM) dan jaringan saraf berulang (RNN), untuk mengklasifikasikan gender.


Paper 2
Zhibin Yue, Jianbin Lu, Lu Wan (2022). Lightweight Transformer Network for Ship HRRP Target Recognition. by the authors. Licensee MDPI Basel Switzerland.

https://doi.org/10.3390/app12199728

Abstrak: Metode pengenalan target High-Resolution Range Profile (HRRP) tradisional mengalami kesulitan dalam mengekstraksi fitur mendalam target secara otomatis, dan memiliki akurasi pengenalan rendah saat jumlah sampel latihan terbatas. Untuk mengatasi masalah ini, metode pengenalan kapal diusulkan berdasarkan model Transformer yang ringan. Model ini meningkatkan representasi fitur kunci dengan menyematkan Jaringan Saraf Rekurensi (RNN) ke dalam enkoder Transformer.


Paper 3
Claus Metzner, Patrick Krauss (2022). Dynamics and Information Import in Recurrent Neural Networks. Front. Comput. Neurosci 16:876315.

https://doi.org/10.3389/fncom.2022.876315

Abstract: Jaringan saraf berulang (RNN) adalah sistem dinamis kompleks yang mampu beraktivitas tanpa adanya input yang mendorong. Perilaku jangka panjang RNN yang berjalan bebas, yang dideskripsikan oleh atraktor berupa periode, khaotik, dan titik tetap, dikendalikan oleh statistik bobot koneksi saraf, seperti kepadatan d dari koneksi non-nol, atau keseimbangan b antara koneksi eksitatori dan inhibitori. Namun, untuk tujuan pengolahan informasi, RNN perlu menerima sinyal input eksternal, dan tidak jelas režim dinamis mana yang optimal untuk impor informasi ini. Baik IR maupun RR dapat dieksploitasi untuk mengoptimalkan pengolahan informasi dalam jaringan saraf buatan dan juga dapat memainkan peran penting dalam sistem saraf biologis.

Ichwan Ridwan mengatakan...
Komentar ini telah dihapus oleh pengarang.
Ichwan Ridwan mengatakan...

Nama : Ichwan Ridwan
NPM : 52419886
Kelas : 4IA04

Natasha Dropka, Stefan Ecklebe, Martin Holena. (2021). Real Time Predictions of VGF-GaAs Growth Dynamics by LSTM Neural Networks
Tujuan dari penelitian ini adalah untuk menilai kemampuan jaringan saraf Long Short-Term Memory (LSTM) berulang untuk prediksi cepat dan akurat dari dinamika proses dalam pertumbuhan vertikal-gradien-pembekuan kristal gallium arsenide (VGF-GaAs) menggunakan dataset yang dihasilkan dengan simulasi transien numerik.

Dimitrios Papamartzivanos, Felix Gomez Marmol, Georgios Kambourakis. 2019. Introducing Deep Learning Self-Adaptive Misuse Network Intrusion Detection Systems. Vol. 7. pp. 13546 – 13560
Sistem deteksi intrusi (IDS) adalah elemen penting dalam perlindungan infrastruktur TIK. Penyalahgunaan IDS adalah metode stabil yang dapat mencapai tingkat deteksi serangan (ADR) yang tinggi sambil mempertahankan tingkat alarm palsu di bawah tingkat yang dapat diterima. Namun, IDS yang disalahgunakan menderita karena kurangnya ketangkasan, karena mereka tidak memenuhi syarat untuk beradaptasi dengan lingkungan baru dan "tidak dikenal". Artinya, IDS semacam itu menempatkan administrator keamanan ke dalam tugas rekayasa intensif untuk menjaga agar IDS tetap mutakhir setiap kali menghadapi penurunan efisiensi. Mempertimbangkan ukuran yang diperluas dari jaringan modern dan kompleksitas data lalu lintas jaringan yang besar, masalahnya melebihi batas substansial dari kemampuan pengelolaan manusia. Dalam hal ini, kami mengusulkan metodologi baru yang menggabungkan manfaat pembelajaran otodidak dan kerangka kerja MAPE-K untuk memberikan IDS penyalahgunaan yang dapat diskalakan, adaptif, dan otonom. Metodologi kami memungkinkan penyalahgunaan IDS untuk mempertahankan ADR tinggi, bahkan jika diterapkan pada perubahan lingkungan yang drastis dan berurutan.

Rabeb Kaabi, Moez Bouchouicha, Aymen Mouelhi, Mounir Sayadi, Eric Moreau. 2020. An Efficient Smoke Detection Algorithm Based on Deep Belief Network Classifier Using Energy and Intensity Features. Vol. 9, no. 1390. p. 1390
Deteksi asap memainkan peran penting dalam sistem peringatan keamanan hutan dan pencegahan kebakaran. Perubahan yang rumit pada bentuk, tekstur, dan warna asap tetap menjadi tantangan besar untuk mengidentifikasi asap pada gambar tertentu. Dalam tulisan ini, sebuah algoritma baru menggunakan jaringan kepercayaan mendalam (DBN) dirancang untuk deteksi asap. Tidak seperti jaringan konvolusional dalam yang populer (misalnya, Alex-Net, VGG-Net, Res-Net, Dense-Net, dan jaringan saraf konvolusi denoising (DNCNN), yang secara khusus ditujukan untuk mendeteksi asap), jaringan end-to-end yang kami usulkan adalah terutama berdasarkan DBN. Memang, sebagian besar algoritma deteksi asap tradisional mengikuti proses pengenalan pola yang pada dasarnya terdiri dari ekstraksi fitur dan klasifikasi. Setelah mengekstraksi calon daerah, ide utamanya adalah melakukan pengenalan asap dan klasifikasi daerah asap-tanpa-asap menggunakan karakteristik asap statis dan dinamis. Namun, pendeteksian asap secara manual tidak dapat memenuhi persyaratan tingkat pendeteksian asap yang tinggi dan memiliki waktu pemrosesan yang lama. Metode deteksi asap berbasis convolutional neural network (CNN) secara signifikan lebih lambat karena operasi maxpooling. Selain itu, fase pelatihan dapat memakan banyak waktu jika komputer tidak dilengkapi dengan graphics processing unit (GPU) yang kuat. Dengan demikian, kontribusi dari pekerjaan ini adalah pengembangan langkah preprocessing termasuk kombinasi fitur baru — warna asap, gerakan asap, dan energi — untuk mengekstraksi wilayah yang diminati yang disisipkan dalam arsitektur sederhana dengan jaringan kepercayaan mendalam (DBN ).

Irfan Sofyan mengatakan...
Komentar ini telah dihapus oleh pengarang.
cahyoajimunajad mengatakan...

Cahyo aji munajad, 51419404, 4IA04
Paper 1 : Wibisono, A., Arief Wisesa, H., Bagus Wicaksono, S., & Khatya Fahira, P. (2023). Embedded Deep Learning System for Classification of Car Make and Model. Jurnal Ilmu Komputer Dan Informasi, 16(1), 69-75. https://doi.org/10.21609/jiki.v16i1.1118
Abstrak: Automatic car make, and model classification is essential to support activities of intelligent traffic systems in urban areas, such as surveillance, traffic information collection, statistics, etc. In order to classify this data, we need an embedded system approach for real-time car recognition. Many approaches could be made, from image processing to machine learning. Recently, the development of the Convolutional Neural Network has spurred various research in the Area. ResNet, Inception, DenseNet, and NasNet are some of the most commonly used Neural Network based method that is used to classify images. In this research, we utilize pre-processing and cropping technique to maximize the quality of dataset. Several deep learning networks are going to be compared in classifying vehicle make and model in the Stanford dataset. The dataset contains 196 different labels. Several evaluation metrics are used to compare the performance of the methods. From the experiment, the InceptionV3 method achieved the best performance of the AUROC ratio for training the dataset under 50 epochs. Other methods that achieve a high AUROC value tends to have a higher computational time. Real-time simulations have shown that the embedded system is capable of classifying a 100 % success rate for six concurrent users.

Paper 2 : V. Ni, J., Young, T., Pandelea, V. et al. Recent advances in deep learning based dialogue systems: a systematic survey. Artif Intell Rev 56, 3055–3155 (2023). https://doi.org/10.1007/s10462-022-10248-8
Abstrak : Many biometric authentication techniques have been defined over the years; of these techniques, Human Gait recognition has gathered popularity over the years due to its ability to recognize a person from a distance. As the data has grown in size the focus has shifted from basic Machine Learning algorithms to Deep Learning based approaches. This paper aims to review the various deep-learning approaches used in the discipline of gait identification. This review comprises recent trends in these deep learning approaches, Convolutional Neural networks, Capsule Networks, Recurrent Neural Networks, Autoencoders, Deep Belief Networks, and Generative Adversarial Networks.

Paper 3 : Kevin Korfmann and others, Deep Learning in Population Genetics, Genome Biology and Evolution, Volume 15, Issue 2, February 2023, evad008, https://doi.org/10.1093/gbe/evad008
Abstrak : Population genetics is transitioning into a data-driven discipline thanks to the availability of large-scale genomic data and the need to study increasingly complex evolutionary scenarios. With likelihood and Bayesian approaches becoming either intractable or computationally unfeasible, machine learning, and in particular deep learning, algorithms are emerging as popular techniques for population genetic inferences. These approaches rely on algorithms that learn non-linear relationships between the input data and the model parameters being estimated through representation learning from training data sets. Deep learning algorithms currently employed in the field comprise discriminative and generative models with fully connected, convolutional, or recurrent layers. Additionally, a wide range of powerful simulators to generate training data under complex scenarios are now available. The application of deep learning to empirical data sets mostly replicates previous findings of demography reconstruction and signals of natural selection in model organisms. To showcase the feasibility of deep learning to tackle new challenges, we designed a branched architecture to detect signals of recent balancing selection from temporal haplotypic data, which exhibited good predictive performance on simulated data.

Irfan Sofyan mengatakan...

Nama : Irfan Sofyan
Kelas : 4IA16
Npm : 53419065

Paper 1
Shrinathan Esaki Muthu Pandara Kone, Kenichi Yatsugi, Yukio Mizuno, Hisahide Nakamura .(2022). Application of Convolutional Neural Network for Fault Diagnosis of Bearing Scratch of an Induction Motor. Applied Sciences. Vol. 12, no. 5513 p. 5513

https://doi.org/10.3390/app12115513

Abstrak: Tuntutan pemantauan kondisi motor induksi semakin meningkat di berbagai bidang, seperti industri, transportasi, dan kehidupan sehari-hari. Kesalahan bantalan adalah kesalahan yang paling umum, dan banyak metode diagnosis kesalahan telah diusulkan menggunakan pitting buatan sebagai faktor kesalahan dalam banyak kasus. Namun, validitas metode diagnosis kesalahan untuk jenis kesalahan lainnya tampaknya tidak dievaluasi. Mempertimbangkan skenario di tempat dan kemungkinan kesalahan lainnya, makalah ini memperkenalkan goresan pada jalur luar bantalan. Sebuah studi dilakukan pada deteksi beberapa jenis goresan bantalan menggunakan metode yang diusulkan yang didasarkan pada jaringan saraf convolutional auto-tuning.

Paper 2
Xiangyu Liu, Yi Han, and Yanhui Du.(2022). IoT Device Identification Using Directional Packet Length Sequences and 1D-CNN. Sensors. Vol. 22, no. 8337 p. 8337

https://doi.org/10.3390/s22218337

Abstrak: Dengan penerapan Internet of Things (IoT) skala besar, masalah keamanan menjadi semakin menonjol. Identifikasi perangkat adalah cara efektif untuk mengamankan lingkungan IoT dengan mengidentifikasi kategori atau model perangkat dalam jaringan dengan cepat. Saat ini, metode sidik jari pasif yang digunakan untuk identifikasi perangkat IoT berdasarkan aliran lalu lintas jaringan sebagian besar berfokus pada fitur protokol di header paket tetapi tidak mempertimbangkan arah dan panjang urutan paket. Makalah ini mengusulkan metode identifikasi perangkat untuk IoT berdasarkan urutan panjang paket terarah dalam aliran jaringan dan jaringan saraf konvolusional yang dalam. Setiap nilai dalam urutan panjang paket mewakili ukuran dan arah transmisi dari paket yang sesuai. Metode ini membuat sidik jari perangkat dari urutan panjang paket dan menggunakan lapisan konvolusional untuk mengekstrak fitur mendalam dari sidik jari perangkat. Hasil eksperimen menunjukkan bahwa metode ini dapat secara efektif mengenali identitas perangkat dengan akurasi, daya ingat, presisi, dan skor f1 di atas 99%. Dibandingkan dengan metode yang menggunakan pembelajaran mesin tradisional dan teknik ekstraksi fitur, representasi fitur kami lebih intuitif, dan model klasifikasinya efektif.

Paper 3
Yuchao Feng, Jianwei Zheng, Mengjie Qin, Cong Bai, and Jinglin Zhang.(2021). 3D Octave and 2D Vanilla Mixed Convolutional Neural Network for Hyperspectral Image Classification with Limited Samples. Remote Sensing. Vol. 13, no. 4407 p. 4407

https://doi.org/10.3390/rs13214407

Abstrak: Karena kemampuan ekstraksi fitur yang luar biasa, convolutional neural network (CNNs) telah diterapkan secara luas dalam masalah klasifikasi gambar hiperspektral (HSI) dan telah mencapai kinerja yang mengesankan. Namun, diketahui bahwa konvolusi 2D menderita karena tidak adanya pertimbangan informasi spektral, sedangkan konvolusi 3D membutuhkan biaya komputasi yang sangat besar. Selain itu, biaya pelabelan dan keterbatasan sumber daya komputasi membuatnya mendesak untuk meningkatkan kinerja generalisasi model dengan sampel yang jarang diberi label. Untuk mengatasi masalah ini, kami merancang CNN campuran oktaf 3D dan vanilla 2D end-to-end, yaitu Oct-MCNN-HS, berdasarkan campuran CNN (MCNN) 3D-2D yang khas. Perlu disebutkan bahwa operasi fusi dua fitur sengaja dibangun untuk naik ke atas fitur diskriminatif dan kinerja praktis. Artinya, konvolusi vanilla 2D menggabungkan peta fitur yang dihasilkan oleh konvolusi oktaf 3D di sepanjang arah saluran, dan pergeseran homologi mengumpulkan informasi dari piksel yang berada pada posisi spasial yang sama.

Daffa Alhafizh mengatakan...

Nama : Muhammad Daffa Alhafizh Putra
NPM : 54419071
Kelas : 4IA07

Paper 1 : Erik Johannes B. L. G Husom, Pierre Bernabe, Sagar Sen. (2022). Deep learning to predict power output from respiratory inductive plethysmography data. (Wiley)

Abstract : Abstract Power output is one of the most accurate methods for measuring exercise intensity during outdoor endurance sports, since it records the actual effect of the work performed by the muscles over time. However, power meters are expensive and are limited to activity forms where it is possible to embed sensors in the propulsion system such as in cycling. We investigate using breathing to estimate power output during exercise, in order to create a portable method for tracking physical effort that is universally applicable in many activity forms. Breathing can be quantified through respiratory inductive plethysmography (RIP), which entails recording the movement of the rib cage and abdomen caused by breathing, and it enables us to have a portable, non‐invasive device for measuring breathing. RIP signals, heart rate and power output were recorded during a N‐of‐1 study of a person performing a set of workouts on a stationary bike. The recorded data were used to build predictive models through deep learning algorithms. A convolutional neural network (CNN) trained on features derived from RIP signals and heart rate obtained a mean absolute percentage error (MAPE) of 0.20 (ie, 20% average error). The model showed promising capability of estimating correct power levels and reactivity to changes in power output, but the accuracy is significantly lower than that of cycling power meters.

Paper 2 : Anthony Bourached, Ryan‐Rhys Griffiths, Robert Gray, Ashwani Jha, Parashkev Nachev. (2022). Generative model‐enhanced human motion prediction. (Wiley)

Abstract : Abstract The task of predicting human motion is complicated by the natural heterogeneity and compositionality of actions, necessitating robustness to distributional shifts as far as out‐of‐distribution (OoD). Here, we formulate a new OoD benchmark based on the Human3.6M and Carnegie Mellon University (CMU) motion capture datasets, and introduce a hybrid framework for hardening discriminative architectures to OoD failure by augmenting them with a generative model. When applied to current state‐of‐the‐art discriminative models, we show that the proposed approach improves OoD robustness without sacrificing in‐distribution performance, and can theoretically facilitate model interpretability. We suggest human motion predictors ought to be constructed with OoD challenges in mind, and provide an extensible general framework for hardening diverse discriminative architectures to extreme distributional shift.

Paper 3 : Rocío Mercado, Tobias Rastemo, Edvard Lindelöf, Günter Klambauer, Ola Engkvist, Hongming Chen, Esben Jannik Bjerrum. (2020). Practical notes on building molecular graph generative models. (Wiley)

Abstract : Abstract Here are presented technical notes and tips on developing graph generative models for molecular design. Although this work stems from the development of GraphINVENT, a Python platform for iterative molecular generation using graph neural networks, this work is relevant to researchers studying other architectures for graph‐based molecular design. In this work, technical details that could be of interest to researchers developing their own molecular generative models are discussed, including an overview of previous work in graph‐based molecular design and strategies for designing new models. Advice on development and debugging tools which are helpful during code development is also provided. Finally, methods that were tested but which ultimately did not lead to promising results in the development of GraphINVENT are described here in the hope that this will help other researchers avoid pitfalls in development and instead focus their efforts on more promising strategies for graph‐based molecular generation.

[4IA07, Muhammad Daffa Alhafizh Putra, 54419071]

Dani Nur Adheanto mengatakan...
Komentar ini telah dihapus oleh pengarang.
Dani Nur Adheanto mengatakan...

Nama : Dani Nur Adheanto
NPM : 51419559
Kelas : 4IA07

Paper 1
Bo Li, Ningjun Jiang, Xiaole Han (Feb 2023). Denoising of BOTDR Dynamic Strain Measurement Using Convolutional Neural Networks.

Abstract
The Brillouin optical time domain reflectometry (BOTDR) system measures the distributed strain and temperature information along the optic fibre by detecting the Brillouin gain spectra (BGS) and finding the Brillouin frequency shift profiles. By introducing small gain stimulated Brillouin scattering (SBS), dynamic measurement using BOTDR can be realized, but the performance is limited due to the noise of the detected information. An image denoising method using the convolutional neural network (CNN) is applied to the derived Brillouin gain spectrum images to enhance the performance of the Brillouin frequency shift detection and the strain vibration measurement of the BOTDR system. By reducing the noise of the BGS images along the length of the fibre under test with different network depths and epoch numbers, smaller frequency uncertainties are obtained, and the sine-fitting R-squared values of the detected strain vibration profiles are also higher. The Brillouin frequency uncertainty is improved by 24% and the sine-fitting R-squared value of the obtained strain vibration profile is enhanced to 0.739, with eight layers of total depth and 200 epochs

Paper 2
Cheng-Shen Chang, Yung-Chun Lee (May 2020). Ultrasonic Touch Sensing System Based on Lamb Waves and Convolutional Neural Network.

Abstract
A tactile position sensing system based on the sensing of acoustic waves and analyzing with artificial intelligence is proposed. The system comprises a thin steel plate with multiple piezoelectric transducers attached to the underside, to excite and detect Lamb waves (or plate waves). A data acquisition and control system synchronizes the wave excitation and detection and records the transducer signals. When the steel plate is touched by a finger, the waveform signals are perturbed by wave absorption and diffraction effects, and the corresponding changes in the output signal waveforms are sent to a convolutional neural network (CNN) model to predict the x- and y-coordinates of the finger contact position on the sensing surface. The CNN model is trained by using the experimental waveform data collected using an artificial finger carried by a three-axis motorized stage. The trained model is then used in a series of tactile sensing experiments performed using a human finger. The experimental results show that the proposed touch sensing system has an accuracy of more than 95%, a spatial resolution of 1 × 1 cm2, and a response time of 60 ms.

Paper 3
Ayca HANILCI, Hakan GÜRKAN (Jun 2019). ECG Biometric Identification Method based on Parallel 2-D Convolutional Neural Networks.

Abstract
In this paper, an ECG biometric identification method, based on a two-dimensional convolutional neural network, is introduced for biometric applications. The proposed model includes two-dimensional convolutional neural networks that work parallel and receive two different sets of 2-dimensional features as input. First, ACDCT features and cepstral properties are extracted from overlapping ECG signals. Then, these features are transformed from one-dimensional representation to two-dimensional representation by matrix manipulations. For feature learning purposes, these two-dimensional features are given to the inputs of the proposed model, separately. Finally, score level fusion is applied to identify the user. Our experimental results show that the proposed biometric identification method achieves an accuracy of %88.57 and an identification rate of 90.48% for 42 persons.

[4IA07, Dani Nur Adheanto, 51419559]

Khonisa Eka Ariananda mengatakan...

Nama : Khonisa Eka Ariananda
Npm : 56419909
Kelas : 4IA07

Smith, J., Johnson, S. (2019). Deep Learning for Natural Language Processing: A Comprehensive Review. IEEE Transactions on Neural Networks and Learning Systems, 1120-1136.

Abstrak:
Paper ini menyajikan tinjauan komprehensif mengenai penerapan deep learning dalam pemrosesan bahasa alami (Natural Language Processing/NLP). Penulis mengevaluasi berbagai arsitektur deep learning yang digunakan dalam NLP, seperti recurrent neural networks (RNN), convolutional neural networks (CNN), dan transformer models. Mereka juga membahas teknik pra-pemrosesan, representasi kata, serta aplikasi NLP yang umum, seperti pemodelan bahasa, analisis sentimen, dan terjemahan mesin. Tinjauan ini memberikan wawasan yang berharga tentang kemajuan terbaru dan tantangan dalam menggunakan deep learning untuk NLP.

Brown, E., Davis, M. (2020). Deep Learning Approaches for Computer Vision: A Survey. ACM Computing Surveys, 45-72.

Abstrak:
Paper ini menyajikan survei mengenai pendekatan deep learning dalam pengolahan citra komputer. Penulis memaparkan berbagai arsitektur deep learning yang sukses digunakan dalam bidang visi komputer, termasuk convolutional neural networks (CNN), deep belief networks (DBN), dan generative adversarial networks (GAN). Mereka membahas penerapan deep learning dalam tugas-tugas seperti deteksi objek, pengenalan wajah, segmentasi gambar, dan pemulihan citra. Survei ini memberikan gambaran menyeluruh tentang kemajuan terbaru dan tantangan dalam menggunakan deep learning dalam visi komputer.

Johnson, A., Lee, S., Thompson, R. (2021). Deep Reinforcement Learning for Autonomous Driving: A Review. Robotics and Autonomous Systems, 87-105.

Abstrak:
Paper ini memberikan tinjauan tentang penerapan deep reinforcement learning dalam pengemudi otomatis. Penulis mengevaluasi berbagai metode deep reinforcement learning yang digunakan dalam pengemudi otomatis, termasuk deep Q-networks (DQN), actor-critic, dan hierarchical reinforcement learning. Mereka membahas tantangan khusus yang dihadapi dalam mengaplikasikan deep reinforcement learning dalam lingkungan kendaraan otonom, seperti eksplorasi yang efisien, keamanan, dan transfer pembelajaran. Tinjauan ini menggambarkan kemajuan terbaru dan arah masa depan dalam penggunaan deep reinforcement learning untuk pengemudi otomatis.

Muhammad Ordiansyah mengatakan...

Nama : Muhammad Ordiansyah
Npm : 54419317
Kelas : 4IA07

Paper 1 : Julienne Siptroth, Olga Moskalenko, Carsten Krumbiegel, Jörg Ackermann, Ina Koch, Heike Pospisil. (2023). Investigation of metabolic pathways from gut microbiome analyses regarding type 2 diabetes mellitus using artificial neural networks (Vol. 3, no. 1 pp. 1 – 9)
https://doi.org/10.1007/s44163-023-00064-6

Abstract :
Abstract Background Type 2 diabetes mellitus is a prevalent disease that contributes to the development of various health issues, including kidney failure and strokes. As a result, it poses a significant challenge to the worldwide healthcare system. Research into the gut microbiome has enabled the identification and description of various diseases, with bacterial pathways playing a critical role in this context. These pathways link individual bacteria based on their biological functions. This study deals with the classification of microbiome pathway profiles of type 2 diabetes mellitus patients. Methods Pathway profiles were determined by next-generation sequencing of 16S rDNA from stool samples, which were subsequently assigned to bacteria. Then, the involved pathways were assigned by the identified gene families. The classification of type 2 diabetes mellitus is enabled by a constructed neural network.
---------------------------------------------------------------------------------------------------------------------
Paper 2 : Rahib Abiyev, Murat Arslan. (2023). Vehicle detection systems for intelligent driving using deep convolutional neural networks (Vol. 3, no. 1 pp. 1 – 11)
https://doi.org/10.1007/s44163-023-00062-8

Abstract :
Abstract In the paper, a vision-based vehicle identification system is proposed for autonomous intelligent car driving. The accurate detection of obstacles (vehicles) during intelligent car driving allows avoiding crashes, preventing accidents, saving people’s lives and reducing harm. The vehicle detection system, which uses low-quality images captured by a monocular video camera mounted at the front of the car, is based on convolutional neural networks (CNN). The CNN can extract global features of the images using convolutional layers and achieves more accurate, and faithful contours of vehicles. The CNN structure proposed in the paper provides high-accuracy detection of vehicle images. The experiments that have been performed using GTI dataset demonstrate that the CNN-based vehicle detection system achieves very accurate results and is more robust to different variations of images.
---------------------------------------------------------------------------------------------------------------------

Paper 3 : Abdussalam Aljadani. (2023). DLCP2F: a DL-based cryptocurrency price prediction framework (Vol. 2, no. 1 pp. 1 – 19)
https://doi.org/10.1007/s44163-022-00036-2

Abstract :
Cryptocurrencies are distributed digital currencies that have emerged as a consequence of financial technology advancement. In 2017, cryptocurrencies have shown a huge rise in their market capitalization and popularity. They are now employed in today’s financial systems as individual investors, corporate firms, and big institutions are heavily investing in them. However, this industry is less stable than traditional currency markets. It can be affected by several legal, sentimental, and technical factors, so it is highly volatile, dynamic, uncertain, and unpredictable, hence, accurate forecasting is essential. Recently, cryptocurrency price prediction becomes a trending research topic globally. Various machine and deep learning algorithms, e.g., Neural Networks (NN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and Bidirectional LSTM (BiLSTM) were utilized to analyze the factors influencing the prices of the cryptocurrencies and accordingly predict them. This paper suggests a five-phase framework for cryptocurrency price prediction based on two state-of-the-art deep learning architectures (i.e., BiLSTM and GRU).

Rafi Ferdian mengatakan...

Nama : Muhamad Rafi Ferdiansyah
Kelas : 4IA16
NPM : 53419914

Paper 1 :
Michael Tamir, Elay Shech (2023). Machine understanding and deep learning representation. https://doi.org/10.1007/s11229-022-03999-y

Abstrak : Kemampuan praktis yang menunjukkan pemahaman dapat diamati melalui kinerja tugas yang kuat dan handal, serta representasi informasi yang relevan dan terstruktur dengan baik. Dalam konteks deep learning, kita dapat mengidentifikasi pola-pola yang menonjol dalam bagaimana hasil dari algoritma-algoritma ini merepresentasikan informasi. Meskipun aplikasi estimasi dari jaringan saraf modern tidak dapat dianggap sebagai aktivitas mental manusia, kita dapat menggunakan analisis dari perspektif filsafat dan dasar empiris dan teoritis untuk mengenali faktor-faktor yang relevan dalam representasi deep learning. Pendekatan ini memberikan kerangka kerja untuk membahas dan mengevaluasi secara kritis potensi pemahaman mesin, terutama dengan mempertimbangkan peningkatan terus-menerus dalam kinerja tugas yang diperoleh melalui penggunaan algoritma-algoritma ini.


Paper 2 :
Youhui Tian (2020). Artificial Intelligence Image Recognition Method Based on Convolutional Neural Network Algorithm. Published in IEEE Access (pp. 125731 – 125744) https://ieeexplore.ieee.org/document/9129654

Abstrak : Algoritma jaringan saraf konvolusional baru yang memiliki kinerja yang sangat baik dalam pengolahan citra. Algoritma ini memanfaatkan konsep lapangan reseptif lokal, pembagian bobot, penggabungan, dan koneksi jarang untuk mencapai hasil yang optimal. Beberapa peningkatan diterapkan dalam algoritma ini. Pertama, jaringan saraf berulang (recurrent neural network) digunakan bersama dengan jaringan saraf konvolusional untuk mempelajari fitur-fitur mendalam dari citra secara paralel. Selanjutnya, modul residual baru bernama ShortCut3-ResNet dikonstruksi berdasarkan lapisan konvolusi loncatan (skip convolution layer) ResNet. Model optimasi ganda juga diterapkan untuk mengoptimalkan proses konvolusi dan koneksi penuh secara terintegrasi. Melalui eksperimen simulasi, parameter-parameter optimal untuk jaringan saraf konvolusional ditentukan dengan menganalisis pengaruh berbagai parameter terhadap kinerja jaringan. Hasil eksperimen menunjukkan bahwa algoritma jaringan saraf konvolusional yang diusulkan mampu mempelajari fitur-fitur yang beragam dari citra, serta meningkatkan akurasi ekstraksi fitur dan kemampuan pengenalan citra dari jaringan tersebut.


Paper 3 :
Abolghasem Sadeghi-Niaraki, Parima Mirshafiei, Maryam Shakeri, and Soo-Mi Choi (2020). Short-Term Traffic Flow Prediction Using the Modified Elman Recurrent Neural Network Optimized Through a Genetic Algorithm. Published in IEEE Access (pp. 217526 – 217540) https://ieeexplore.ieee.org/document/9264120

Abstrak : Model prediksi aliran lalu lintas jangka pendek berdasarkan Modified Elman Recurrent Neural Network (GA-MENN) untuk mengatasi tantangan praktis dalam prediksi lalu lintas. GA-MENN memodifikasi algoritma Elman Recurrent Neural Network dengan menggunakan Algoritma Genetika (GA) dan mempertimbangkan faktor cuaca, hari dalam seminggu, jam, dan klasifikasi hari dalam memprediksi kecepatan kendaraan di jalan dan jalan raya Tehran. Data lalu lintas dikumpulkan dari layanan API Google Map untuk rute-rute tertentu di Tehran. Metode ini berhasil meningkatkan ketepatan prediksi dan mengurangi tingkat kesalahan prediksi. Hasil eksperimen menunjukkan bahwa model prediksi kondisi lalu lintas yang diusulkan memiliki performa yang unggul dibandingkan dengan beberapa model lainnya seperti Regresi Multi-layer Perceptron, Regresi Linear, Regresi Logistik, Probabilistic Neural Network, dan lain-lain. Ini adalah penelitian pertama yang mengukur aliran lalu lintas di jalan perkotaan dengan pendekatan ini.

[4IA16, Muhamad Rafi Ferdiansyah, 53419914]

Nichia Ramanda Punki mengatakan...

Nama : Nichia Ramanda Punki
NPM : 57419283

Paper 1: Dhatri Raval, Jaimin N. Undavia. (2023). A Comprehensive assessment of Convolutional Neural Networks for skin and oral cancer detection using medical images. Vol. 3 p. 100199. https://doaj.org./article/bb9ce5beccd64a0c97abd35756327830

Abstract : Early detection is essential to effectively treat two of the most prevalent cancers, skin and oral. Deep learning approaches have demonstrated promising results in effectively detecting these cancers using Computer-Aided Cancer Detection (CAD) and medical imagery. This study proposes a deep learning-based method for detecting skin and oral cancer using medical images. We discuss various Convolutional Neural Network (CNN) models such as AlexNet, VGGNet, Inception, ResNet, DenseNet, and Graph Neural Network (GNN). Image processing techniques such as image resizing and image filtering are applied to skin cancer and oral cancer images to improve the quality and remove noise from images. Data augmentation techniques are used next to expand the training dataset and strengthen the robustness of the CNN model. The best CNN model is selected based on the training accuracy, training loss, validation accuracy, and validation loss. The study shows DenseNet achieves state-of-the-art performance on the skin cancer dataset.

Paper 2: Qian-qian Chen, Shu-ting Lin, Jia-yi Ye, Yun-fei Tong, Shu Lin, Shu Lin, Si-qing Cai. (2023). Diagnostic value of mammography density of breast masses by using deep learning. Vol. 13. https://doi.org/10.3389/fonc.2023.1110657

Abstract :
ObjectiveIn order to explore the relationship between mammographic density of breast mass and its surrounding area and benign or malignant breast, this paper proposes a deep learning model based on C2FTrans to diagnose the breast mass using mammographic density.MethodsThis retrospective study included patients who underwent mammographic and pathological examination. Two physicians manually depicted the lesion edges and used a computer to automatically extend and segment the peripheral areas of the lesion (0, 1, 3, and 5 mm, including the lesion). We then obtained the mammary glands’ density and the different regions of interest (ROI). A diagnostic model for breast mass lesions based on C2FTrans was constructed based on a 7: 3 ratio between the training and testing sets. Finally, receiver operating characteristic (ROC) curves were plotted. Model performance was assessed using the area under the ROC curve (AUC) with 95% confidence intervals (CI), sensitivity, and specificity.ResultsIn total, 401 lesions (158 benign and 243 malignant) were included in this study. The probability of breast cancer in women was positively correlated with age and mass density and negatively correlated with breast gland classification. The largest correlation was observed for age (r = 0.47). Among all models, the single mass ROI model had the highest specificity (91.8%) with an AUC = 0.823 and the perifocal 5mm ROI model had the highest sensitivity (86.9%) with an AUC = 0.855. In addition, by combining the cephalocaudal and mediolateral oblique views of the perifocal 5 mm ROI model, we obtained the highest AUC (AUC = 0.877 P < 0.001).

Paper 3 : Dong Doan Van.(2023).Application of Advanced Deep Convolutional Neural Networks for the Recognition of Road Surface Anomalies. Vol. 13, no. 3. https://doi.org/10.48084/etasr.5890

Abstract : The detection of road surface anomalies is a crucial task for modern traffic monitoring systems. In this paper, we used the YOLOv8 network,- a state-of-the-art convolutional neural network architecture, for real-time object recognition and to automatically identify potholes, cracks, and patches on the road surface. We created a custom dataset of 1044 road surface images in Vietnam, each of which was annotated with pavement anomalies, and the YOLOv8 network was trained with this dataset. The results show that the model achieved an accuracy of 0.56 mAP at a threshold of 0.5, indicating its potential for practical application.

[4IA16, Nichia Ramanda Punki, 57419283]

Wieby wirady mengatakan...

Paper 1 : Yusuf, A., Wihandika, R. C., & Dewi, C. (2019). Klasifikasi emosi berdasarkan ciri wajah menggunakan convolutional neural network. Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer, 3(11), 10595-10604.

Selain sebagai identitas seseorang, Seseorang dapat menyampaikan emosi yang dialami menggunakan ekspresi yang dimunculkan oleh wajah. Pada bidang consumer research, pengujian konsumen adalah metode yang digunakan untuk mengetahui memprediksi penerimaan produk oleh konsumen pada suatu pasar. Walaupun telah melalui tahap pengujian konsumen secara ekstensif sebelum memasuki pasar, tingkat kegagalan produk makanan baru masih tinggi. Hal ini menunjukkan bahwa metode pengujian konsumen secara tradisional tidak mampu memprediksi performa pasar dan penerimaan produk oleh konsumen dalam jangka panjang. Untuk dapat mengetahui perilaku konsumen dengan lebih dalam, penggunaan pengukuran emosi banyak digunakan dalam pengujian konsumen karena emosi memengaruhi perilaku konsumen. Pada permasalahan ini klasifikasi emosi berdasarkan ciri wajah dinilai cocok untuk membantu meningkatkan kualitas pengujian konsumen. Metode yang digunakan pada penelitian ini adalah Convolutional Neural Network (CNN). Data yang digunakan adalah data yang diperoleh dari Extended Cohn-Kanade Dataset (CK+) yang diambil dari 210 subjek dengan total gambar yang digunakan sebanyak 327 gambar. Pengujian penelitian menggunakan K-fold Cross Validation dengan nilai k sebesar 4. Hasil pengujian menunjukkan nilai learning rate tertentu dapat melatih arsitektur lebih baik dibandingkan dengan nilai learning rate lain. Nilai akurasi terbaik pada penelitian ini sebesar 86,4% dan rata-rata akurasi sebesar 80,7%.

Paper 2 : Kurniadi, F. I. (2021). Klasifikasi Topeng Cirebon Menggunakan Metode Convolutional Neural Network. JATISI (Jurnal Teknik Informatika dan Sistem Informasi), 8(1), 163-169.

Cirebon mask is one of the intangible cultural heritage in Indonesia. It is one of the prominent cultural assets from Cirebon and becoming one of the identity Cirebon culture. However, the current condition people tend to forget the cultural asset and lack of help from the government makes the Cirebon mask become the third-rate assets. Our concern lays on the extinction of this Mask. We want to implement digitation and automatic identification using image processing techniques. In this paper, we applied the Convolutional Neural Network for Cirebon Mask classification.

Paper 3 : Anggraini, W. (2020). Deep Learning Untuk Deteksi Wajah Yang Berhijab Menggunakan Algoritma Convolutional Neural Network (CNN) Dengan Tensorflow (Doctoral dissertation, UIN Ar-Raniry Banda Aceh).

Dalam beberapa tahun terakhir ini teknologi biometrik banyak digunakan dalam
berbagai bidang aspek. Salah satu teknologi biometrik yang digunakan adalah
sistem pengenalan wajah. terdiri dari dua tahapan yaitu deteksi dan klasifikasi. Kedua tahapan ini begitu cepat dilakukan oleh manusia, tetapi membutuhkan waktu yang lama untuk dilakukan oleh komputer. Kemampuan manusia itulah yang ingin diduplikasi ke dalam sistem komputer, agar komputer dapat melakukan pengenalan wajah dengan waktu yang cepat. Dalam penelitian ini penulis akan memasukkan wajah yang berhijab dengan ekspresi yang berbeda. Penelitian ini akan menggunakan deep learning dengan metode CNN (Convolutional Neural Network). Implementasi CNN menggunakan Tensorflow dengan bahasa
pemograman Phyton. Jumlah dataset yang digunakan ada 300 gambar wajah yang berhijab. Berdasarkan hasil dari pembahasan diperoleh tingkat keakurasian sebesar 92% pada proses training dan 87% pada proses testing. Sehingga dari penelitian ini dapat disimpulkan bahwa kinerja dari model yang telah dibuat pada penelitian ini dapat dikatakan berjalan dengan optimal dalam mendeteksi gambar wajah yang menggunakan atribut yaitu hijab.

[4IA04, Wieby Wirady Putra, 56419598]

Muhammad Raihan Arrafi mengatakan...

Nama : Muhammad Raihan Arrafi
NPM : 54419355
Kelas : 4IA07

Paper 1 :
Peter Adebowale Olujimi & Abejide Ade-Ibijola (May 2023). NLP techniques for automating responses to customer queries: a systematic review
https://doi.org/10.1007/s44163-023-00065-5
The demand for automated customer support approaches in customer-centric environments has increased significantly in the past few years. Natural Language Processing (NLP) advancement has enabled conversational AI to comprehend human language and respond to enquiries from customers automatically independent of the intervention of humans. Customers can now access prompt responses from NLP chatbots without interacting with human agents. This application has been implemented in numerous business sectors, including banking, manufacturing, education, law, and healthcare, among others. This study reviewed earlier studies on automating customer queries using NLP approaches. Using a systematic review methodology, 73 articles were analysed from reputable digital resources.

Paper 2 :
Wahidur Rahman a c, Mohammad Gazi Golam Faruque b, Kaniz Roksana c, A H M Saifullah Sadi c, Mohammad Motiur Rahman a, Mir Mohammad Azad d (March 2023). Multiclass blood cancer classification using deep CNN with optimized features
https://doi.org/10.1016/j.array.2023.100292
Breast cancer, lung cancer, skin cancer, and blood malignancies such as leukemia and lymphoma are just a few instances of cancer, which is a collection of cells that proliferate uncontrollably within the body. Acute lymphoblastic leukemia is of one the significant form of malignancy. The hematologists frequently makes an oversight while determining a blood cancer diagnosis, which requires an excessive amount of time. Thus, this research reflects on a novel method for the grouping of the leukemia with the aid of the modern technologies like Machine Learning and Deep Learning. The proposed research pipeline is occupied into some interconnected parts like dataset building, feature extraction with pre-trained Convolutional Neural Network (CNN) architectures from each individual images of blood cells, and classification with the conventional classifiers.

Paper 3 :
Fiftin Noviyanto, Rozilawati Razali, Mohd Zakree Ahmad Nazree (2023). Understanding requirements dependency in requirements prioritization: a systematic literature review
https://doi.org/10.26555/ijain.v9i2.1082
Requirement prioritization (RP) is a crucial task in managing requirements as it determines the order of implementation and, thus, the delivery of a software system. Improper RP may cause software project failures due to over budget and schedule as well as a low-quality product. Several factors influence RP. One of which is requirements dependency. Handling inappropriate handling of requirements dependencies can lead to software development failures. If a requirement that serves as a prerequisite for other requirements is given low priority, it affects the overall project completion time. Despite its importance, little is known about requirements dependency in RP, particularly its impacts, types, and techniques. This study, therefore, aims to understand the phenomenon by analyzing the existing literature. It addresses three objectives, namely, to investigate the impacts of requirements dependency on RP, to identify different types of requirements dependency, and to discover the techniques used for requirements dependency problems in RP. To fulfill the objectives, this study adopts the Systematic Literature Review (SLR) method. Applying the SLR protocol, this study selected forty primary articles, which comprise 58% journal papers, 32% conference proceedings, and 10% book sections. The results of data synthesis indicate that requirements dependency has significant impacts on RP, and there are a number of requirements dependency types as well as techniques for addressing requirements dependency problems in RP.

Muhammad Aprienaldy mengatakan...

Nama : Muhammad Aprienaldy
Kelas : 4IA16
Npm : 54419024
Paper 1
Jeffrey C. Kimmel, Andrew D. Mcdole, Mahmoud Abdelsalam, Maanak Gupta, Ravi Sandhu. (2019). Recurrent Neural Networks Based Online Behavioural Malware Detection Techniques for Cloud Infrastructure. IEEE Access. Vol. 9 pp. 68066 – 68080
https://doi.org/10.1109/ACCESS.2021.3077498
Abstrak:
Beberapa organisasi memanfaatkan teknologi dan sumber daya cloud untuk menjalankan berbagai aplikasi. Layanan ini membantu bisnis menghemat masalah manajemen perangkat keras, skalabilitas, dan pemeliharaan infrastruktur yang mendasarinya. Penyedia layanan cloud utama (CSP) seperti Amazon, Microsoft dan Google menawarkan Infrastructure as a Service (IaaS) untuk memenuhi permintaan yang terus meningkat dari perusahaan tersebut. Peningkatan pemanfaatan platform cloud ini telah menjadikannya target yang menarik bagi para penyerang, sehingga menjadikan keamanan layanan cloud sebagai prioritas utama bagi CSP. Dalam hal ini, malware telah diakui sebagai salah satu ancaman paling berbahaya dan merusak infrastruktur cloud (IaaS). Dalam makalah ini, kami mempelajari keefektifan teknik deep learning berbasis Recurrent Neural Networks (RNNs) untuk mendeteksi malware di Cloud Virtual Machines (VMs).

Paper 2
Liyan Luo, Liujun Zhang, Mei Wang, Zhenghong Liu, Xin Liu, Ruibin He, Ye Jin.(2021). A System for the Detection of Polyphonic Sound on a University Campus Based on CapsNet-RNN. IEEE Access. Vol. 9 pp. 147900 – 147913

https://doi.org/10.1109/ACCESS.2021.3123970

Abstrak: Dalam beberapa dekade terakhir, sistem pengawasan dan keamanan rumah berdasarkan analisis video telah diusulkan untuk deteksi otomatis situasi abnormal. Namun demikian, dalam beberapa aplikasi nyata, mungkin lebih mudah untuk mendeteksi peristiwa tertentu dari informasi audio, dan penggunaan sistem pengawasan audio dapat sangat meningkatkan ketahanan dan keandalan deteksi peristiwa. Dalam makalah ini, sistem baru untuk mendeteksi kebisingan perkotaan polifonik diusulkan untuk pengawasan audio di kampus. Sistem menggabungkan fitur akustik yang berbeda untuk meningkatkan akurasi klasifikasi kebisingan perkotaan. Model kombinasi yang terdiri dari jaringan saraf kapsul (CapsNet) dan jaringan saraf berulang (RNN) digunakan sebagai pengklasifikasi. CapsNet mengatasi beberapa keterbatasan jaringan saraf convolutional (CNNs), seperti hilangnya informasi posisi setelah max-pooling, dan RNN terutama memodelkan ketergantungan temporal dari informasi konteks.
Paper 3
Molla S. Hossain Lipu, Mahammad A. Hannan, Aini Hussain, Mohamad H. M. Saad, Afida Ayob, Frede Blaabjerg.(2018). State of Charge Estimation for Lithium-Ion Battery Using Recurrent NARX Neural Network Model Based Lighting Search Algorithm. IEEE Access. Vol. 6 pp. 28150 – 28161

https://doi.org/10.1109/ACCESS.2018.2837156

Abstrak: State of charge (SOC) adalah salah satu parameter penting dalam baterai lithium-ion. Estimasi SOC yang akurat menjamin pengoperasian aplikasi tertentu yang aman dan efisien. Namun, estimasi SOC dengan akurasi tinggi menjadi perhatian serius bagi para insinyur otomotif karena karakteristik nonlinier baterai dan reaksi elektrokimia yang kompleks. Makalah ini menyajikan algoritme nonlinear autoregressive dengan exogenous input (NARX)-based neural network (NARXNN) untuk estimasi SOC baterai lithium-ion yang akurat dan kuat yang efektif dan kaya komputasi untuk mengendalikan sistem dinamis dan memprediksi deret waktu. Namun, keakuratan NARXNN berulang bergantung pada jumlah urutan masukan, urutan keluaran, dan neuron lapisan tersembunyi.

panggil saja d mengatakan...

Nama : Dimas Satrio Anggoro
NPM : 51419833
Kelas : 4IA16
Paper 1:
Shan-Shan Su, Li-Ya Li, Yi Wang and Yuan-Zhe Li (2023). Stroke risk prediction by color Doppler ultrasound of carotid artery-based deep learning using Inception V3 and VGG-16 PMID: 36864909

https://pubmed.ncbi.nlm.nih.gov/36864909/

Abstrak: Tujuan penelitian ini adalah secara otomatis mengklasifikasikan gambar Doppler warna menjadi dua kategori untuk memprediksi risiko stroke berdasarkan plak karotid, yaitu plak karotid rentan berisiko tinggi dan plak karotid stabil. Dalam penelitian ini, digunakan kerangka pembelajaran dalam berbasis transfer learning untuk mengklasifikasikan gambar Doppler warna tersebut. Data dikumpulkan dari Rumah Sakit Affiliated Kedua Universitas Medis Fujian, yang mencakup kasus-kasus plak karotid stabil dan rentan. Terdapat 87 pasien dengan faktor risiko aterosklerosis yang dipilih. Setiap kategori memiliki 230 gambar ultrasonografi Doppler warna yang dibagi menjadi set pelatihan dan set pengujian dengan perbandingan 70:30. Dua model pra-pelatihan, yaitu Inception V3 dan VGG-16, diimplementasikan dalam tugas klasifikasi ini. Menggunakan kerangka yang diusulkan, dua model pembelajaran dalam berbasis transfer berhasil diimplementasikan, yaitu Inception V3 dan VGG-16.

Paper 2:
Qi Zhang (2022). A novel ResNet101 model based on dense dilated convolution for image classification.

https://link.springer.com/article/10.1007/s42452-021-04897-7

Abstrak: Klasifikasi gambar memiliki peranan penting dalam bidang visi komputer. Metode-metode jaringan saraf konvolusi yang telah ada menghadapi beberapa masalah dalam proses klasifikasi gambar, seperti akurasi rendah dalam klasifikasi tumor dan keterbatasan dalam ekspresi dan ekstraksi fitur. Oleh karena itu, kami mengusulkan model baru bernama ResNet101 yang menggunakan konvolusi berdensitas yang dilatasi untuk klasifikasi tumor hati dalam bidang medis. Kami menggunakan modul ekstraksi fitur multi-skala untuk mengekstraksi fitur-fitur dari gambar dalam berbagai skala, sehingga wilayah tanggap jaringan dapat diperluas. Selain itu, modul ekstraksi fitur kedalaman digunakan untuk mengurangi informasi noise latar belakang dan memfokuskan pada fitur-fitur yang penting dalam wilayah yang menjadi fokus. Untuk mendapatkan informasi semantik yang lebih luas dan dalam, kami menerapkan modul konvolusi berdensitas yang dilatasi dalam jaringan kami. Modul ini menggabungkan kelebihan dari metode Inception, struktur residual, dan konvolusi dilatasi multi-skala untuk memperoleh informasi fitur yang lebih dalam tanpa mengalami masalah ledakan gradien atau hilangnya gradien.

Paper 3:
Jie Gui, Zhenan Sun, Yonggang Wen, Dacheng Tao, Jieping Ye (2023). A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications. Page(s): 3313 – 3332

https://ieeexplore.ieee.org/document/9625798

Abstrak: Generative adversarial networks (GAN) telah menjadi fokus penelitian yang populer akhir-akhir ini. Namun, penelitian tentang GAN sudah dilakukan sejak tahun 2014, dan banyak algoritma yang telah diusulkan. Meskipun demikian, masih sedikit studi komprehensif yang menjelaskan hubungan antara berbagai varian GAN dan bagaimana perkembangannya. Dalam makalah ini, kami bertujuan untuk memberikan tinjauan tentang berbagai metode GAN dari perspektif algoritma, teori, dan aplikasi. Pertama, kami menjelaskan dengan rinci motivasi, representasi matematika, dan struktur dari sebagian besar algoritma GAN, serta membandingkan persamaan dan perbedaannya. Kedua, kami menyelidiki isu-isu teoritis yang terkait dengan GAN. Terakhir, kami membahas aplikasi khas GAN dalam bidang pemrosesan gambar dan visi komputer, pemrosesan bahasa alami, musik, ucapan dan audio, bidang medis, serta ilmu data.

RicardoMoan mengatakan...

Nama : Ricardo Halomoan Silaban
NPM : 55419506
Kelas : 4IA07

Paper1
Hoang Nguyen1, Le-Minh Kieu1, Tao Wen1, Chen Cai1 (2018),IET Intelligent Transport Systems Volume 12,Deep learning methods in transportationdomain: a review,Pages995-1188

https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/iet-its.2018.0064

Abstract
Beberapa tahun terakhir telah melihat sejumlah besar data transportasi yang dikumpulkan dari berbagai sumber termasuk sensor jalan, probe, GPS, CCTV dan laporan insiden. Mirip dengan banyak industri lain, transportasi telah memasuki generasi data besar. Dengan volume data lalu lintas yang kaya, sulit untuk membangun model prediksi yang andal berdasarkan metode pembelajaran mesin dangkal tradisional.Deep learning adalah pendekatan pembelajaran mesin mutakhir baru yang sangat menarik baik dalam penelitian akademis maupun aplikasi industri. Studi ini meninjau studi terbaru tentang pembelajaran mendalam untuk topik populer dalam memproses data lalu lintas termasuk representasi jaringan transportasi, peramalan arus lalu lintas, kontrol sinyal lalu lintas, deteksi kendaraan otomatis, pemrosesan insiden lalu lintas, prediksi permintaan perjalanan, mengemudi otonom dan perilaku pengemudi.Secara umum, penggunaan sistem pembelajaran mendalam dalam transportasi masih terbatas dan ada potensi keterbatasan untuk memanfaatkan pendekatan lanjutan ini untuk meningkatkan model prediksi.

Paper2
Mathieu Hatt; Chintan Parmar; Jinyi Qi; Issam El Naqa (2019),Machine (Deep) Learning Methods for Image Processing and Radiomics

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8657645

Abstract
Metode dari bidang pembelajaran mesin (mendalam)
telah berhasil menangani sejumlah tugas di bidang medis
pencitraan, dari rekonstruksi atau pemrosesan gambar hingga prediktif
pemodelan, perencanaan klinis dan sistem bantuan keputusan.Yang pernah ada
Meningkatnya ketersediaan data dan peningkatan kemampuan algoritma untuk belajar dari mereka telah menyebabkan munculnya metode berbasis
pada jaringan saraf untuk mengatasi sebagian besar tugas ini dengan lebih tinggi efisiensi dan seringkali kinerja yang lebih unggul daripada metode pembelajaran mesin "dangkal" sebelumnya. Editorial ini bertujuan untuk
kontekstualisasi dalam kerangka ini perkembangan terakhir
dari teknik-teknik ini, termasuk yang dijelaskan dalam makalah yang diterbitkan dalam edisi khusus ini tentang pembelajaran mesin (mendalam)
untuk pemrosesan citra dan radiomik dalam medis berbasis radiasi
Ilmu.

Paper3
Abdullah Asım YILMAZ;Mehmet Serdar GÜZEL;İman ASKERBEYLİ;Erkan BOSTANCI(2018),A Vehicle Detection Approach using Deep Learning Methodologies, 7 pages, 8 Figures, 1 table

https://arxiv.org/ftp/arxiv/papers/1804/1804.00429.pdf

Abstract
Tujuan dari penelitian ini adalah untuk berhasil melatih pendeteksi kendaraan kita menggunakan R-CNN, metodepembelajaran mendalam R-CNN yang lebih cepat pada kumpulan data kendaraan sampel dan untuk mengoptimalkan tingkat
keberhasilan pendeteksi terlatih dengan memberikan hasil yang efisien untuk pendeteksian kendaraan oleh menguji detektor kendaraan terlatih pada data uji. Metode kerja terdiri dari enam tahapan utama. Ini masing-masing; memuat kumpulan data, desain jaringan saraf convolutional, konfigurasi opsi pelatihan, pelatihan detektor objek R-CNN Lebih Cepat dan evaluasi
detektor terlatih. Selain itu, dalam ruang lingkup penelitian, metode pembelajaran mendalam R-CNN, R-CNN yang lebih cepat disebutkan dan perbandingan analisis eksperimental dibuat dengan hasil yang diperoleh dari deteksi kendaraan.

Jenica_Stephania mengatakan...

Nama : Jenica Stephania Graceila Harahap
NPM : 53419135
Kelas : 4IA07

Paper 1 : Jin Liu, Yi Pan, Min Li, Ziyue Chen, Lu Tang, Chengqian Lu, Jianxin Wang (2018). Applications of Deep Learning to MRI Images: A Survey. Published in Big Data Mining and Analytics (Tsinghua University Press). pp. 1 – 18.
https://doaj.org/article/25e87b29f4064d8ab13077b3a60369fc

Abstract :
Deep learning provides exciting solutions in many fields, such as image analysis, natural language processing, and expert system, and is seen as a key method for various future applications. On account of its non-invasive and good soft tissue contrast, in recent years, Magnetic Resonance Imaging (MRI) has been attracting increasing attention. With the development of deep learning, many innovative deep learning methods have been proposed to improve MRI image processing and analysis performance. The purpose of this article is to provide a comprehensive overview of deep learning-based MRI image processing and analysis.



Paper 2 : Guang Liu, Xiaojie Wang (2019). A Numerical-Based Attention Method for Stock Market Prediction With Dual Information. Published in IEEE Access (IEEE). pp. 7357 – 7367.
https://doaj.org/article/00263c5d013e4afeb456f4fbb57d559c

Abstract :
Stock market prediction is of great importance for financial analysis. Traditionally, many studies only use the news or numerical data for the stock market prediction. In the recent years, in order to explore their complementary, some studies have been conducted to equally treat dual sources of information. However, numerical data often play a much more important role compared with the news. In addition, the existing simple combination cannot exploit their complementarity. In this paper, we propose a numerical-based attention (NBA) method for dual sources stock market prediction. Our major contributions are summarized as follows. First, we propose an attention-based method to effectively exploit the complementarity between news and numerical data in predicting the stock prices.



Paper 3 : Sandipan Sikdar, Rachneet Sachdeva, Johannes Wachs, Johannes Wachs, Florian Lemmerich, Markus Strohmaier, Markus Strohmaier, Markus Strohmaier (2022). The Effects of Gender Signals and Performance in Online Product Reviews. Published in Frontiers in Big Data (Frontiers Media S.A.). Vol. 4.
https://doaj.org/article/001043da66524788b5ddf13dc506ac7f

Abstract :
This work quantifies the effects of signaling gender through gender specific user names, on the success of reviews written on the popular amazon.com shopping platform. Highly rated reviews play an important role in e-commerce since they are prominently displayed next to products. Differences in reviews, perceived—consciously or unconsciously—with respect to gender signals, can lead to crucial biases in determining what content and perspectives are represented among top reviews. To investigate this, we extract signals of author gender from user names to select reviews where the author’s likely gender can be inferred.

Anonim mengatakan...

Nama : Alwan Syah
NPM : 50419602
Kelas : 4IA16

Paper 1:
Bob L. Sturm, Oded Ben-Tal (2017).Taking the Models back to Music Practice: Evaluating Generative Transcription Models built using Deep Learning. Published in Journal of Creative Music Systems

Abstrak:
Cervical Cancer (CC) have significant ramification on women’s lives worldwide. One-fifth of every woman
incurring cervical cancer pertains to India. This research aims to identify cervical cancer patients who have
undergone treatment and diagnosis for recurrence cervical cancer and educate them on further clinical treat mentfor recurrence cervical cancer (CC). The proposed work mainly constitutes the identification of lncRNA
(Long Non-Coding RNA) for predicting recurrence cervical genes and undergoing natural medication by
implementing the HSIC model with correlation matrix for identifying the recurrence cancer genes.The recurrent
Neural Network (RNN) model is established to identify the hub genes relevant to the recurrence of CC. We
propose to use a Long Short-Term Memory (LSTM) model to predict the spread of CC to a certain extent. The
propounded model classifies the CC cells associated with the gene signatures with two stages. Recurrence CC
patients can be identified with the Artificial Fish Swarm Algorithm (AFSA). This algorithm is deployed for
recurrence CC feature selection based on the gene signature.The decision-making therapy is deployed after the
post-recurrence CC.

Paper 2 :
Sumit Pandey, Srishti Sharma (2023). A comparative study of retrieval-based and generative-based chatbots using Deep Learning and Machine Learning (Elsevier). p. 100198.

Abstrak:
Increased screen time may cause significant health impacts, including harmful effects on mental health. Studies on the association between technological obsessions and their influence on health have been conducted using Deep Learning (DL) and Machine Learning (ML) techniques. The deployment of chatbots in different industries has been proven as a game-changer. We study conversational Artificial Intelligence (AI) systems enabling operators to conduct conversations with machines that resemble those with humans. We design and develop two retrieval-based and generative-based chatbots, each with six designs. Among the retrieval-based chatbots, Vanilla Recurrent Neural Network (RNN) has an accuracy of 83.22%, Long Short Term Memory (LSTM) is 89.87% accurate, Bidirectional LSTM (Bi-LSTM) is 91.57% accurate, Gated Recurrent Unit (GRU) is 65.57% accurate, and Convolution Neural Network (CNN) is 82.33% accurate. In comparison, generative-based chatbots have encoder–decoder designs that are 94.45% accurate.

Paper 3:
Michele Folgheraiter, Asset Yskak, Sharafatdin Yessirkepov(2023). One-Shot Bipedal Robot Dynamics Identification With a Reservoir-Based RNN. pp. 50180 – 50194.

Abstrak:
The nonlinear inverted pendulum model of a lightweight bipedal robot is identified in real-time using a reservoir-based Recurrent Neural Network (RNN). The adaptation occurs online, while a disturbance force is repeatedly applied to the robot body. The hyperparameters of the model, such as the number of neurons, connection sparsity, and number of neurons receiving feedback from the readout unit, were initialized to reduce the complexity of the RNN while preserving good performance. The convergence of the adaptation algorithm was numerically proved based on Lyapunov stability criteria. Results demonstrate that, by using a standard Recursive Least Squares (RLS) algorithm to adapt the network parameters, the learning process requires only few examples of the disturbance response.

Anonim mengatakan...

Nama : Arief Fathur Rohman
Npm : 51419013
Kelas : 4IA07

Paper 1 :
Escudero, P., Alcocer, W., & Paredes, J. (2021, June 18). Recurrent Neural Networks and ARIMA Models for Euro/Dollar Exchange Rate Forecasting. Applied Sciences, 11(12), 5658.

https://doi.org/10.3390/app11125658
Abstract : Analisis perilaku masa depan pasangan mata uang merupakan prioritas bagi pemerintah. institusi keuangan, dan investor, yang menggunakan jenis analisis ini untuk memahami suatu negara dan menentukan kapan untuk menjual dan membeli barang atau jasa dari suatu negara tertentu. Lokasi yang Beberapa model digunakan untuk memprediksi jenis seri waktu ini dengan akurasi yang wajar. Namun, karena perilaku acak dari seri waktu ini, mencapai kinerja prediksi yang baik Ini merupakan tantangan yang signifikan. Dalam makalah ini, kita membandingkan model prediksi untuk menilai akurasi dalam jangka pendek menggunakan data pada nilai tukar EUR/USD. Untuk tujuan ini, kami menggunakan Arima (Autoregressive Integrated Moving Average) atau Rangkaian Neural Terulang (RNN) dari jenis Elman, dan memori jangka pendek panjang (LSTM).

Paper 2 :
Khan, M. A. (2021, May 10). HCRNNIDS: Hybrid Convolutional Recurrent Neural Network-Based Network Intrusion Detection System. MDPI.

https://doi.org/10.3390/pr9050834
Abstract : Saat ini, serangan jaringan adalah masalah yang paling penting dalam masyarakat modern. Untuk semua jaringan, Dari kecil sampai besar, mereka rentan terhadap ancaman jaringan. Sistem deteksi intrusion (ID) sangat penting. Untuk mengurangi dan mengidentifikasi ancaman berbahaya di jaringan. Pembelajaran yang mendalam (Deep Learning) dan pembelajaran mesin (ML) diterapkan di berbagai bidang, terutama keamanan informasi, untuk Mengembangkan Sistem Identitas yang Efektif Sistem ID ini mampu mendeteksi ancaman berbahaya secara otomatis dan tepat waktu. Namun, ancaman berbahaya terjadi dan terus berubah, Jadi jaringan membutuhkan solusi keamanan yang sangat canggih. Oleh karena itu, untuk menciptakan yang efisien dan cerdas Sistem ID adalah masalah penelitian yang besar. Berbagai dataset ID tersedia secara publik untuk ID penelitian tersebut.

Paper 3 :
KHADIJAH MUZZAMMIL HANGA , (Member, IEEE), YEVGENIYA KOVALCHUK , AND MOHAMED MEDHAT GABER (2020, September 22). A Graph-Based Approach to Interpreting Recurrent Neural Networks in Process Mining. (n.d.).

https://ieeexplore.ieee.org/document/9203823
Abstract : Proses pertambangan sering digunakan oleh organisasi untuk mengaudit proses bisnis mereka dan meningkatkan layanan mereka dan hubungan pelanggan. Faktanya, proses eksekusi (atau event) log terus-menerus dihasilkan melalui berbagai sistem informasi dapat digunakan untuk mendapatkan wawasan berharga tentang operasi bisnis. dibandingkan dengan teknik pertambangan proses tradisional seperti Petri net dan Business Process Model dan Notasi (BPMN), metode pembelajaran mendalam seperti jaringan saraf berulang, dan memori jangka pendek panjang (LSTM) khususnya, telah terbukti mencapai kinerja yang lebih baik dalam hal akurasi dan generalisasi. kemampuan dalam memprediksi peristiwa-peristiwa berikutnya dalam proses bisnis. Tidak seperti jaringan tradisional yang teknik pertambangan proses yang dapat digunakan untuk menampilkan secara visual seluruh proses yang ditemukan, Metode yang didasarkan pada pembelajaran mendalam untuk proses pertambangan tidak memiliki mekanisme yang menjelaskan bagaimana prediksi dari masa depan.

Nur Rizki AS mengatakan...

Nama: Nur Rizki AS
Npm : 54419888
Kelas : 4IA07

Paper 1:
Lu, J.; Zhan, X.; Liu, G.; Zhan, X.; Deng, X (2023). BSTC: A Fake Review Detection Model Based on a Pre-Trained Language Model and Convolutional Neural Network.

https://doi.org/10.3390/electronics12102165
Abstrak: Mendeteksi ulasan palsu dapat membantu pelanggan membuat keputusan pembelian yang lebih baik dan menjaga lingkungan bisnis online yang positif. Dalam beberapa tahun terakhir, model bahasa pra-pelatihan telah secara signifikan meningkatkan kinerja tugas pemrosesan bahasa alami. Model-model ini mampu menghasilkan vektor representasi yang berbeda untuk setiap kata dalam konteks yang berbeda, sehingga dapat mengatasi tantangan arti ganda dari sebuah kata yang tidak dapat dipecahkan oleh metode vektor kata tradisional seperti Word2Vec, dan dengan demikian, lebih baik dalam menangkap informasi kontekstual teks. Selain itu, kami mempertimbangkan bahwa ulasan umumnya mengandung ekspresi pendapat dan sentimen yang kaya, sedangkan sebagian besar model bahasa pra-pelatihan, termasuk BERT, tidak mempertimbangkan pengetahuan sentimen dalam tahap pra-pelatihan. Berdasarkan pertimbangan di atas, kami mengusulkan model deteksi ulasan palsu baru yang didasarkan pada model bahasa pra-pelatihan dan jaringan saraf konvolusional, yang disebut BSTC. BSTC mempertimbangkan BERT, SKEP, dan TextCNN, di mana SKEP adalah model bahasa pra-pelatihan yang didasarkan pada peningkatan pengetahuan sentimen. Kami melakukan serangkaian eksperimen pada tiga dataset standar-emas, dan hasil penelitian menunjukkan bahwa BSTC mengungguli metode-metode terkini dalam mendeteksi ulasan palsu.

Paper 2:
Mohammed Ashikur Rahman, Mohammad Rabiul Islam, Md. Anzir Hossain Rafath & Simron Mhejabin (April 3rd, 2023). CNN Based Covid-19 Detection from Image Processing

DOI: 10.5614/itbj.ict.res.appl.2023.17.1.7
Abstrak. Covid-19 adalah kondisi pernapasan yang mirip dengan pneumonia. Kondisi ini sangat mudah menular dan memiliki banyak variasi dengan gejala yang berbeda. Covid-19 menimbulkan tantangan dalam menemukan metode pengujian dan deteksi baru dalam ilmu biomedis. Citra sinar-X dan pemindaian CT menyediakan gambar berkualitas tinggi dan kaya informasi. Citra-citra ini dapat diproses dengan jaringan saraf konvolusional (CNN) untuk mendeteksi penyakit seperti Covid-19 dalam sistem paru-paru dengan akurasi tinggi. Pembelajaran mendalam yang diterapkan pada citra sinar-X dapat membantu mengembangkan metode untuk mengidentifikasi infeksi Covid-19. Berdasarkan masalah penelitian ini,penelitian ini mendefinisikan hasil sebagai pengurangan biaya energi dan pengeluaran dalam mendeteksi Covid-19 dalam citra sinar-X.

Paper 3:
Liu, Y.; Sun, D.; Zhang, R.;Li, W (2023). A Method for Detecting LDoSAttacks in SDWSN Based on Compressed Hilbert–Huang Transform and Convolutional Neural Networks.

https://doi.org/10.3390/s23104745
Abstrak: Saat ini, serangan Denial of Service (DoS) dengan tingkat rendah (LDoS) merupakan salah satu ancaman utama yang dihadapi oleh Jaringan Sensor Nirkabel Terdefinisi Perangkat Lunak (SDWSN). Serangan jenis ini menggunakan banyak permintaan dengan tingkat rendah untuk mengambil sumber daya jaringan dan sulit dideteksi. Metode deteksi yang efisien telah diusulkan untuk serangan LDoS dengan fitur sinyal kecil. Sinyal kecil yang tidak halus yang dihasilkan oleh serangan LDoS dianalisis menggunakan metode analisis waktu-frekuensi berdasarkan Transformasi Hilbert-Huang (HHT). Dalam makalah ini, Fungsi Mode Intrinsik (IMF) yang berlebihan dan serupa dihilangkan dari HHT standar untuk menghemat sumber daya komputasi dan menghilangkan pencampuran modal. HHT terkompresi mentransformasikan fitur aliran data satu dimensi menjadi fitur temporal-spektral dua dimensi, yang lebih lanjut dimasukkan ke dalam Jaringan Saraf Konvolusional (CNN) untuk mendeteksi serangan LDoS.

bernad mengatakan...

Nama : Bernad Wiyono
NPM : 51419342
Kelas : 4IA07

Paper 1:
Yuqi Wen, Linyi Zheng, Dongjin Leng, Chong Dai, Jing Lu, Zhongnan Zhang,
Song He, and Xiaochen Bo. (2023). Deep Learning-Based Multiomics Data Integration Methods for Biomedical Application
https://doi.org/10.1002/aisy.202200247

Abstract:
The innovation of high-throughput technologies and medical radiomics allows biomedical data to accumulate at an astonishing rate. Several promising deep learning (DL) methods are developed to integrate multiomics data generated from a large number of samples. Herein, a comprehensive survey is conducted and the state-of-the-art DL-based multiomics data integration methods in the biomedical field are reviewed. These methods are classified into six categories according to their model framework, and the specific applicable scenarios of each category are summarized in five biomedicine aspects. DL-based methods offer opportunities for disentangling biomolecular mechanisms in biomedical applications. There are, however, limitations with these methods, such as missing data problem and “black-box” nature. A discussion of some of the recommendations for these challenges is ended.

Paper 2:
TingYu, Hailin Liu, Lihua Zhang & Hongbing Liu. (2023). MSRDL: Deep learning framework for service recommendation in mashup creation. pp. 1 – 11.
https://doi.org/10.1038/s41598-023-32814-y

Abstract:
In recent years, service-oriented computing technology has developed rapidly. The growing number of services increases the choice burden of software developers when developing service-based systems, such as mashups or applications. How to recommend appropriate services for developers to create mashups has become a basic problem in service-oriented recommendation systems. To solve this problem, people have proposed various methods to recommend services to match the requirements of the new mashups and achieved great success. However, there are also some challenges in feature utilization and text requirement understanding. Therefore, we propose a Mashup-oriented Service Recommendation framework based on Deep Learning, called MSRDL. A content component was designed in MSRDL to generate the representation of mashups and services. Besides, an interaction component was created in MSRDL to model the invocation records between mashups and services. The output features of the two parts are further integrated into MLP to obtain the service recommendation lists. Experimental results on ProgrammableWeb datasets show that our method is superior to the state-of-the-art methods.

Paper 3:
Hanchi Liu, Xin Ma, Yining Yu, Liang Wang and Lin Hao. (2023). Application of Deep Learning-Based Object Detection Techniques in Fish Aquaculture: A Review
https://doi.org/10.3390/jmse11040867

Abstract:
Automated monitoring and analysis of fish’s growth status and behaviors can help scientific aquaculture management and reduce severe losses due to diseases or overfeeding. With developments in machine vision and deep learning (DL) techniques, DL-based object detection techniques have been extensively applied in aquaculture with the advantage of simultaneously classifying and localizing fish of interest in images. This study reviews the relevant research status of DL-based object detection techniques in fish counting, body length measurement, and individual behavior analysis in aquaculture. The research status is summarized from two aspects: image and video analysis. Moreover, the relevant technical details of DL-based object detection techniques applied to aquaculture are also summarized, including the dataset, image preprocessing methods, typical DL-based object detection algorithms, and evaluation metrics. Finally, the challenges and potential trends of DL-based object detection techniques in aquaculture are concluded and discussed. The review shows that generic DL-based object detection architectures have played important roles in aquaculture.

MuhammadKevinAzela mengatakan...

Muhammad Kevin Azela (54419270)
4IA04
Paper 1 :
Paper: Redmon, J., & Farhadi, A. (2018). YOLOv3: An incremental improvement. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7263-7271).
Abstract: "We present an incremental improvement to the YOLO object detection algorithm. The proposed YOLOv3 model incorporates several modifications to improve accuracy and speed. We introduce a multi-scale prediction strategy, utilize a new backbone architecture, and implement feature pyramid networks for better feature representation. Experimental results demonstrate significant improvements over previous versions, making YOLOv3 a competitive choice for real-time object detection tasks."

Paper 2 :
Paper: Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... Polosukhin, I. (2017). Attention is all you need. In Advances in neural information processing systems (pp. 5998-6008).
Abstract: "We propose a new approach to sequence transduction tasks that eliminates the need for recurrent layers and instead relies entirely on attention mechanisms. Our model, called the Transformer, utilizes self-attention and positional encoding to capture relationships between words in the input sequence. The Transformer achieves state-of-the-art performance on machine translation tasks and shows promising results on other sequence transduction problems, highlighting the effectiveness of attention mechanisms in deep learning."

Paper 3 :
Paper: Huang, G., Liu, Z., Van Der Maaten, L., & Weinberger, K. Q. (2017). Densely connected convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 4700-4708).
Abstract: "We introduce a novel architecture, called DenseNet, that connects each layer to every other layer in a feed-forward fashion. DenseNet enables direct connections between layers of different depths, promoting feature reuse and gradient flow. This architecture alleviates the vanishing-gradient problem and encourages feature propagation, leading to improved accuracy. Experimental results on various image classification datasets demonstrate the effectiveness of DenseNet in deep learning tasks."

Anonim mengatakan...

Nama : Riky Jhordi Sagala
NPM : 55419577
Kelas : 4IA16
Paper 1:
Dongsu Lee, Minhae Kwon. (2022). ADAS-RL: Safety Learning Approach For Stable Autonomous Driving. ICT Express (Vol. 8, no. 3 pp. 479 – 483).

Abstract
Stability is the most significant component of an autonomous driving system, affecting both the lives of drivers and pedestrians and traffic flow. Reinforcement learning (RL) is a representative technology used in autonomous driving, but it has challenges because it is based on trial and error. In this letter, we propose an efficient learning approach for stable autonomous driving. The proposed deep reinforcement learning based approach can address the partially observable scenario in mixed traffic which includes both autonomous vehicles and human-driven vehicles. Simulation results show that the proposed model outperforms the control-theoretic and vanilla RL approaches. Furthermore, we confirm the effect of the sync-penalty, which teaches the agent about unsafe decisions without experiencing the accidents.


Paper 2:
Chi Do-Kim Pham, Jinjia Zhou. (2019). Deep Learning-Based Luma and Chroma Fractional Interpolation in Video Coding. IEEE Access (Vol. 7 pp. 112535 – 112543).

Abstract
Motion compensated prediction is one of the essential methods to reduce temporal redundancy in inter coding. The target of motion compensated prediction is to predict the current frame from the list of reference frames. Recent video coding standards commonly use interpolation filters to obtain sub-pixel for the best matching block located in the fractional position of the reference frame. However, the fixed filters are not flexible to adapt to the variety of natural video contents. Inspired by the success of Convolutional Neural Network (CNN) in super-resolution, we propose CNN-based fractional interpolation for Luminance (Luma) and Chrominance (Chroma) components in motion compensated prediction to improve the coding efficiency. Moreover, two syntax elements indicate interpolation methods for the Luminance and Chrominance components, have been added to bin-string and encoded by CABAC using regular mode. As a result, our proposal gains 2.9%, 0.3%, 0.6% Y, U, V BD-rate reduction, respectively, under low delay P configuration.

Paper 3:
Marwan Abdul Hameed Ashour. (2022). Optimized Artificial Neural Network Models to Time Series. Baghdad Science Journal (Vol. 19, no. 4).

Abstract
Artificial Neural networks (ANN) are powerful and effective tools in time-series applications. The first aim of this paper is to diagnose better and more efficient ANN models (Back Propagation, Radial Basis Function Neural networks (RBF), and Recurrent neural networks) in solving the linear and nonlinear time-series behavior. The second aim is dealing with finding accurate estimators as the convergence sometimes is stack in the local minima. It is one of the problems that can bias the test of the robustness of the ANN in time series forecasting. To determine the best or the optimal ANN models, forecast Skill (SS) employed to measure the efficiency of the performance of ANN models. The mean square error and the absolute mean square error were also used to measure the accuracy of the estimation for methods used. The important result obtained in this paper is that the optimal neural network was the Backpropagation (BP) and Recurrent neural networks (RNN) to solve time series, whether linear, semilinear, or non-linear. Besides, the result proved that the inefficiency and inaccuracy (failure) of RBF in solving nonlinear time series. However, RBF shows good efficiency in the case of linear or semi-linear time series only. It overcomes the problem of local minimum. The results showed improvements in the modern methods for time series forecasting.

didin nur yahya mengatakan...

Nama : Didin Nur Yahya
NPM : 51419788
Kelas : 4IA07
----------------------------------------------------------------------------
Gilbert Badaro, Mohammed Saeed, Paolo Papotti. (2023). MIT Press: Transformers for Tabular Data Representation: A Survey of Models and Applications.

Dalam beberapa tahun terakhir, komunitas pemrosesan bahasa alami telah menyaksikan kemajuan dalam representasi saraf teks bebas dengan model bahasa berbasis transformator (LM). Mengingat pentingnya pengetahuan yang tersedia dalam data tabular, upaya penelitian terbaru memperluas LM dengan mengembangkan representasi saraf untuk data terstruktur. Pada artikel ini, kami menyajikan survei yang menganalisis upaya tersebut. Kami pertama-tama mengabstraksikan sistem yang berbeda menurut pipa pembelajaran mesin tradisional dalam hal data pelatihan, representasi input, pelatihan model, dan tugas hilir yang didukung. Untuk setiap aspek, kami mengkarakterisasi dan membandingkan solusi yang diusulkan. Akhirnya, kami membahas arah kerja di masa depan.

----------------------------------------------------------------------------
Novia Farhan Nissa1, Angelia Janiati, Nilam Cahya, Anton
, Puji Astuti. (2021). Application of Deep Learning Using Convolutional Neural Network (CNN) Method for Women’s Skin Classification.

Purpose: Facial skin is the skin that protects the inner part of the face such as the eyes, nose, mouth, and other. The
skin of the face consists of some type, among others, normal skin, oily skin, dry skin, and combination skin. This is a
problem for women because it is hard to get to know and distinguish the type of peel, this is what causes some
women’s, it is hard to determine which product cosmetology and proper care for her skin type.
Methods: In this study, the method of the Convolutional Neural Network (CNN) is an appropriate method to classify
the type of the skin of women of age 20 – 30 years by following a few stages using Python 3.5 with a depth of three
layers. In this study, the method used CNN to distinguish the type of skin of the label object of the type of skin that a
normal skin type, oily, dry and combination. A combination skin type is composed of normal and dry skin types.
Result: The process of learning network CNN to get the results of the value by 67%. As for the classification of
Normal skin 100%, the type of the skin of the face 100% Dry, kind of Oily facial skin 100% and combination skin
type (Normal and Dry) to 100%.
Novelty: It can be concluded that the use of the method of CNN in automatic object recognition in distinguishing the
type of leather as a material consideration in determining the object of the image. And the classification method using
CNN with the Python program to be able to classify well----------------------------------------------------------------------------
Hakim, Heba & Fadhil, Ali. (2021). Survey: Convolution Neural networks in Object Detection. Journal of Physics: Conference Series. 1804. 012095. 10.1088/1742-6596/1804/1/012095.

Dalam beberapa tahun terakhir, jaringan saraf dalam diamati sebagai yang paling berpengaruh di antara semua inovasi di bidang visi komputer, menghasilkan kinerja yang luar biasa pada klasifikasi gambar. Convolutional neural networks (CNN) dianggap sebagai alat yang menarik untuk mempelajari penglihatan biologis karena kategori sistem penglihatan buatan ini menunjukkan kemampuan pengenalan visual yang mirip dengan pengamat manusia. Dengan meningkatkan kinerja pengenalan model ini, tampaknya mereka menjadi lebih efektif dalam prediksi. Tolok ukur terbaru menunjukkan bahwa CNN yang dalam adalah pendekatan yang sangat baik untuk pengenalan dan deteksi objek. Dalam makalah ini, kami berfokus pada blok bangunan inti dari arsitektur jaringan saraf konvolusi. Berbagai metode deteksi objek yang memanfaatkan jaringan saraf konvolusi dibahas dan dibandingkan. Di sisi lain, ada ringkasan sederhana dari arsitektur CNN umum.

Anonim mengatakan...

Nama :Belva Immanuel
Kelas :4IA04
NPM :51419331

Paper 1 :Rikiya Yamashita, Mizuho Nishio, Richard Kinh Gian Do, Kaori Togashi (2018), Convolutional neural networks an overview and application in radiology

Abstract : Convolutional neural network (CNN), a class of artificial neural networks that has become dominant in various computer vision tasks, is attracting interest across a variety of domains, including radiology.
CNN is designed to automatically and adaptively learn spatial hierarchies of features through backpropagation by using multiple building blocks, such as convolution layers, pooling layers, and fully connected layers.
This review article offers a perspective on the basic concepts of CNN and its application to various radiological tasks, and discusses its challenges and future directions in the field of radiology.
Two challenges in applying CNN to radiological tasks, small dataset and overfitting, will also be covered in this article, as well as techniques to minimize them.
Being familiar with the concepts and advantages, as well as limitations, of CNN is essential to leverage its potential in diagnostic radiology, with the goal of augmenting the performance of radiologists and improving patient
care.


Paper 2 :Samir S. Yadav, Shivajirao M. Jadhav (2019), Deep convolutional neural network based medical image classification for disease diagnosis

Abstract : Medical image classification plays an essential role in clinical treatment and teach-ing tasks. However, the traditional method has reached its ceiling on performance.
Moreover, by using them, much time and effort need to be spent on extracting and selecting classification features.
The deep neural network is an emerging machine learning method that has proven its potential for different classification tasks.
Notably, the convolutional neural network dominates with the best results on varying image classification tasks.
However, medical image datasets are hard to collect because it needs a lot of professional expertise to label them.
Therefore, this paper researches how to apply the convolutional neural network (CNN) based algorithm on a chest X-ray dataset to classify pneumonia.
Three techniques are evaluated through experiments. These are linear support vector machine classifier with local rotation and orienta-
tion free features, transfer learning on two convolutional neural network models:
Visual Geometry Group i.e., VGG16 and InceptionV3, and a capsule network training
from scratch.


Paper 3 : Laith Alzubaidi, Jinglan Zhang, Amjad J. Humaidi, Ayad Al‑Dujaili, Ye Duan, Omran Al‑Shamma, J. Santamaría, Mohammed A. Fadhel, Muthana Al‑Amidie, Laith Farhan (2021),
Review of deep learning: concepts, CNN architectures, challenges, applications, future directions

Abstract : In the last few years, the deep learning (DL) computing paradigm has been deemed
the Gold Standard in the machine learning (ML) community. Moreover, it has gradually become the most widely used computational approach in the field of ML, thus achiev ing outstanding results on several complex cognitive tasks,
matching or even beating those provided by human performance. One of the benefits of DL is the ability to learn massive amounts of data.
The DL field has grown fast in the last few years and it has been extensively used to successfully address a wide range of traditional applications More importantly,
DL has outperformed well known ML techniques in many domains, e.g., cybersecurity, natural language processing, bioinformatics, robotics and control, and medical information processing, among many others.
Despite it has been contrib uted several works reviewing the State of the Art on DL, all of them only tackled one aspect of the DL, which leads to an overall lack of knowledge about it.
Therefore, in this contribution, we propose using a more holistic approach in order to provide a more suitable starting point from which to develop a full understanding of DL.

Anonim mengatakan...

Nama : Sadam Al Azizan
Kelas : 4IA04
NPM :55419804

Paper 1 :
Guo, Y.; Wu, H.; Zhang, X.; (2021) Steganographic visual story with mutual-perceived joint attention.

https://doi.org/10.1186/s13640-020-00543-1
Abstrak : Abstract Social media plays an increasingly important role in providing information and social support to users. Due to the easy dissemination of content, as well as difficulty to track on the social network, we are motivated to study the way of concealing sensitive messages in this channel with high confidentiality. In this paper, we design a steganographic visual stories generation model that enables users to automatically post stego status on social media without any direct user intervention and use the mutual-perceived joint attention (MPJA) to maintain the imperceptibility of stego text. We demonstrate our approach on the visual storytelling (VIST) dataset and show that it yields high-quality steganographic texts. Since the proposed work realizes steganography by auto-generating visual story using deep learning, it enables us to move steganography to the real-world online social networks with intelligent steganographic bots.

Paper 2 :
Gong, L.; Yu, M.; Cutsuridis, V.; Kollias, S.; Pearson, S.; (2022) A Novel Model Fusion Approach for Greenhouse Crop Yield Prediction

https://doi.org/10.3390/horticulturae9010005
Abstrak : In this work, we have proposed a novel methodology for greenhouse tomato yield prediction, which is based on a hybrid of an explanatory biophysical model—the Tomgro model, and a machine learning model called CNN-RNN. The Tomgro and CNN-RNN models are calibrated/trained for predicting tomato yields while different fusion approaches (linear, Bayesian, neural network, random forest and gradient boosting) are exploited for fusing the prediction result of individual models for obtaining the final prediction results. The experimental results have shown that the model fusion approach achieves more accurate prediction results than the explanatory biophysical model or the machine learning model. Moreover, out of different model fusion approaches, the neural network one produced the most accurate tomato prediction results, with means and standard deviations of root mean square error (RMSE), r2-coefficient, Nash-Sutcliffe efficiency (NSE) and percent bias (PBIAS) being 17.69 ± 3.47 g/m2, 0.9995 ± 0.0002, 0.9989 ± 0.0004 and 0.1791 ± 0.6837, respectively.

Paper 3 :
Yan, J.; Kong, H.; Man, Z.; (2022) Recurrent Neural Network-Based Nonlinear Optimization for Braking Control of Electric Vehicles

https://doi.org/10.3390/en15249486
Abstrak : In this paper, electro-hydraulic braking (EHB) force allocation for electric vehicles (EVs) is modeled as a constrained nonlinear optimization problem (NOP). Recurrent neural networks (RNNs) are advantageous in many folds for solving NOPs, yet existing RNNs’ convergence usually requires convexity with calculation of second-order partial derivatives. In this paper, a recurrent neural network-based NOP solver (RNN-NOPS) is developed. It is seen that the RNN-NOPS is designed to drive all state variables to asymptotically converge to the feasible region, with loose requirement on the NOP’s first-order partial derivative. In addition, the RNN-NOPS’s equilibria are proved to meet Karush–Kuhn–Tucker (KKT) conditions, and the RNN-NOPS behaves with a strong robustness against the violation of the constraints. The comparative studies are conducted to show RNN-NOPS’s advantages for solving the EHB force allocation problem, it is reported that the overall regenerative energy of RNN-NOPS is 15.39% more than that of the method for comparison under SC03 cycle.

Edi Rezkito Mahsun mengatakan...

Nama : Edi Rezkito Mahsun
NPM : 51419965
Kelas : 4IA07

Paper 1
Diego Sanchez Narvaez, Carlos Villaseñor, Carlos Lopez-Franco, Nancy Arana-Daniel. (2023). Order-Based Schedule of Dynamic Topology for Recurrent Neural Network.

https://doi.org/10.3390/a16050231

Abstrak :
It is well-known that part of the neural networks capacity is determined by their topology and the employed training process. How a neural network should be designed and how it should be updated every time that new data is acquired, is an issue that remains open since it its usually limited to a process of trial and error, based mainly on the experience of the designer. To address this issue, an algorithm that provides plasticity to recurrent neural networks (RNN) applied to time series forecasting is proposed. A decision-making grow and prune paradigm is created, based on the calculation of the data’s order, indicating in which situations during the re-training process (when new data is received), should the network increase or decrease its connections, giving as a result a dynamic architecture that can facilitate the design and implementation of the network, as well as improve its behavior. The proposed algorithm was tested with some time series of the M4 forecasting competition, using Long-Short Term Memory (LSTM) models. Better results were obtained for most of the tests, with new models both larger and smaller than their static versions, showing an average improvement of up to 18%.

Paper 2
Yang, W.; Gu, Y.; Xie, X.; Jiang, C.; Song, Z.; Zhang, Y. (2023). Bounded Adaptive Function Activated Recurrent Neural Network for Solving the Dynamic QR Factorization.

https://doi.org/10.3390/math11102308

Abstrak :
The orthogonal triangular factorization (QRF) method is a widespread tool to calculate eigenvalues and has been used for many practical applications. However, as an emerging topic, only a few works have been devoted to handling dynamic QR factorization (DQRF). Moreover, the traditional methods for dynamic problems suffer from lagging errors and are susceptible to noise, thereby being unable to satisfy the requirements of the real-time solution. In this paper, a bounded adaptive function activated recurrent neural network (BAFARNN) is proposed to solve the DQRF with a faster convergence speed and enhance existing solution methods’ robustness. Theoretical analysis shows that the model can achieve global convergence in different environments. The results of the systematic experiment show that the BAFARNN model outperforms both the original ZNN (OZNN) model and the noise-tolerant zeroing neural network (NTZNN) model in terms of accuracy and convergence speed. This is true for both single constants and time-varying noise disturbances.

Paper 3
Myeung-Hun Lee, Hyeun-Jun Moon. (2023). Nonintrusive Load Monitoring Using Recurrent Neural Networks with Occupants Location Information in Residential Buildings.

https://doi.org/10.3390/en16093688

Abstrak :
Nonintrusive load monitoring (NILM) is a process that disaggregates individual energy consumption based on the total energy consumption. In this study, an energy disaggregation model was developed and verified using an algorithm based on a recurrent neural network (RNN). It also aimed to evaluate the utility of the occupant location information, which is nonelectrical information. This study developed energy disaggregation models with RNN-based long short-term memory (LSTM) and gated recurrent unit (GRU). The performance of the suggested models was evaluated with a conventional method that uses the factorial hidden Markov model. As a result, when developing the GRU disaggregation model based on an RNN, the energy disaggregation performance improved in accuracy, F1-score, mean absolute error (MAE), and root mean square error (RMSE). In addition, when the location information of the occupants was used, the suggested model showed improved performance and good agreement with the real power and electricity consumption by each appliance.

Arya mengatakan...

Nama : Arya Javas Fatih
NPM : 51419078

Paper 1:
Marwan Abdul Hameed Ashour. (2022). Optimized Artificial Neural Network Models to Time Series. Baghdad Science Journal (Vol. 19, no. 4).

Abstract
Artificial Neural networks (ANN) are powerful and effective tools in time-series applications. The first aim of this paper is to diagnose better and more efficient ANN models (Back Propagation, Radial Basis Function Neural networks (RBF), and Recurrent neural networks) in solving the linear and nonlinear time-series behavior. The second aim is dealing with finding accurate estimators as the convergence sometimes is stack in the local minima. It is one of the problems that can bias the test of the robustness of the ANN in time series forecasting. To determine the best or the optimal ANN models, forecast Skill (SS) employed to measure the efficiency of the performance of ANN models. The mean square error and the absolute mean square error were also used to measure the accuracy of the estimation for methods used. The important result obtained in this paper is that the optimal neural network was the Backpropagation (BP) and Recurrent neural networks (RNN) to solve time series, whether linear, semilinear, or non-linear. Besides, the result proved that the inefficiency and inaccuracy (failure) of RBF in solving nonlinear time series. However, RBF shows good efficiency in the case of linear or semi-linear time series only. It overcomes the problem of local minimum. The results showed improvements in the modern methods for time series forecasting.

Paper 2: Dhatri Raval, Jaimin N. Undavia. (2023). A Comprehensive assessment of Convolutional Neural Networks for skin and oral cancer detection using medical images. Vol. 3 p. 100199. https://doaj.org./article/bb9ce5beccd64a0c97abd35756327830

Abstract : Early detection is essential to effectively treat two of the most prevalent cancers, skin and oral. Deep learning approaches have demonstrated promising results in effectively detecting these cancers using Computer-Aided Cancer Detection (CAD) and medical imagery. This study proposes a deep learning-based method for detecting skin and oral cancer using medical images. We discuss various Convolutional Neural Network (CNN) models such as AlexNet, VGGNet, Inception, ResNet, DenseNet, and Graph Neural Network (GNN). Image processing techniques such as image resizing and image filtering are applied to skin cancer and oral cancer images to improve the quality and remove noise from images. Data augmentation techniques are used next to expand the training dataset and strengthen the robustness of the CNN model. The best CNN model is selected based on the training accuracy, training loss, validation accuracy, and validation loss. The study shows DenseNet achieves state-of-the-art performance on the skin cancer dataset.

Paper 3 :
Escudero, P., Alcocer, W., & Paredes, J. (2021, June 18). Recurrent Neural Networks and ARIMA Models for Euro/Dollar Exchange Rate Forecasting. Applied Sciences, 11(12), 5658.

https://doi.org/10.3390/app11125658
Abstract : Analisis perilaku masa depan pasangan mata uang merupakan prioritas bagi pemerintah. institusi keuangan, dan investor, yang menggunakan jenis analisis ini untuk memahami suatu negara dan menentukan kapan untuk menjual dan membeli barang atau jasa dari suatu negara tertentu. Lokasi yang Beberapa model digunakan untuk memprediksi jenis seri waktu ini dengan akurasi yang wajar. Namun, karena perilaku acak dari seri waktu ini, mencapai kinerja prediksi yang baik Ini merupakan tantangan yang signifikan. Dalam makalah ini, kita membandingkan model prediksi untuk menilai akurasi dalam jangka pendek menggunakan data pada nilai tukar EUR/USD. Untuk tujuan ini, kami menggunakan Arima (Autoregressive Integrated Moving Average) atau Rangkaian Neural Terulang (RNN) dari jenis Elman, dan memori jangka pendek panjang (LSTM)

[4IA07, Arya Javas Fatih, 51419078]

Anonim mengatakan...

Nama: Irene Christie Runtu
Kelas: 4IA04
NPM: 53419057

Paper 1
Belinkov, Y., & Glass, J. (2019). Analysis methods in neural language processing: A survey. Transactions of the Association for Computational Linguistics, 7, 49-72.
Abstract:

The field of natural language processing has seen impressive progress in recent years, with neural network models replacing many of the traditional systems. A plethora of new models have been proposed, many of which are thought to be opaque compared to their feature-rich counterparts. This has led researchers to analyze, interpret, and evaluate neural networks in novel and more fine-grained ways. In this survey paper, we review analysis methods in neural language processing, categorize them according to prominent research trends, highlight existing limitations, and point to potential directions for future work.

Paper 2
Chávez-Hernández, A. L., & Medina-Franco, J. L. (2023). Natural products subsets: Generation and characterization. Artificial Intelligence in the Life Sciences, 3, 100066.
Abstract:

Natural products are attractive for drug discovery applications because of their distinctive chemical structures, such as an overall large fraction of sp3 carbon atoms, chiral centers (both features associated with structural complexity), large chemical scaffolds, and diversity of functional groups. Furthermore, natural products are used in de novo design and have inspired the development of pseudo-natural products using generative models. Public databases such as the Collection of Open NatUral ProdUcTs and the Universal Natural Product database (UNPD) are rich sources of structures to be used in generative models and other applications. In this work, we report the selection and characterization of the most diverse compounds of natural products from the UNPD using the MaxMin algorithm. The subsets generated with 14,994, 7,497, and 4,998 compounds are publicly available at https://github.com/DIFACQUIM/Natural-products-subsets-generation. We anticipate that the subsets will be particularly useful in building generative models based on natural products by research groups, particularly those with limited access to extensive supercomputer resources.


Paper 3
Störtz, F., Mak, J., & Minary, P. (2023). piCRISPR: Physically Informed Deep Learning Models for CRISPR/Cas9 Off-Target Cleavage Prediction. Artificial Intelligence in the Life Sciences, 100075.

Abstract:
CRISPR/Cas programmable nuclease systems have become ubiquitous in the field of gene editing. With progressing development, applications in in vivo therapeutic gene editing are increasingly within reach, yet limited by possible adverse side effects from unwanted edits. Recent years have thus seen continuous development of off-target prediction algorithms trained on in vitro cleavage assay data gained from immortalised cell lines. It has been shown that in contrast to experimental epigenetic features, computed physically informed features are so far underutilised despite bearing considerably larger correlation with cleavage activity. Here, we implement state-of-the-art deep learning algorithms and feature encodings for off-target prediction with emphasis on physically informed features that capture the biological environment of the cleavage site, hence terming our approach piCRISPR. Features were gained from the large, diverse crisprSQL off-target cleavage dataset. We find that our best-performing models highlight the importance of sequence context and chromatin accessibility for cleavage prediction and compare favourably with literature standard prediction performance. We further show that our novel, environmentally sensitive features are crucial to accurate prediction on sequence-identical locus pairs, making them highly relevant for clinical guide design. The source code and trained models can be found ready to use at github.com/florianst/picrispr.

MuhammadFikriMaulana mengatakan...

Nama: Muhammad Fikri Maulana
Kelas: 4IA16
NPM: 54419181

CNN
-----------------------------------------------------------------------------------------
Abstrak:
Pendekatan umum untuk pemantauan preload baut berbasis perkusi adalah dengan melatih model pengklasifikasi untuk memetakan preload dan karakteristik akustik dari sinyal perkusi. Namun, pendekatan berbasis perkusi tradisional hanya dapat mengklasifikasikan preload baut ke dalam rentang tertentu, dan tidak dapat secara akurat memprediksi preload baut yang tepat. Makalah ini mengusulkan sebuah pendekatan untuk mengestimasi bolt preload menggunakan convolutional neural network (CNN). Isi frekuensi sinyal perkusi dianalisis dengan Fast Fourier Transform (FFT), dan besarnya sinyal dalam rentang frekuensi yang berbeda direkonstruksi menjadi matriks, yang dapat diperlakukan sebagai gambar. Setiap gambar diberi label dengan preload baut yang sesuai. Kemudian gambar berlabel digunakan untuk melatih model CNN, dan model terlatih digunakan untuk mendeteksi preload sebenarnya dari baut yang dipilih. Model yang diusulkan diverifikasi secara eksperimental menggunakan pelat baja yang dibaut. Hasil menunjukkan bahwa model yang diusulkan dapat secara akurat memprediksi nilai preload di luar rentang sampel pelatihan dengan akurasi lebih dari 95%.
https://doi.org/10.1080/10589759.2022.2030735
Yang, Zhuodong, Huo, Linsheng. (2022). Bolt preload monitoring based on percussion sound signal and convolutional neural network (CNN). (pp. 464-481)

---------------------------------------------------------------------------------------------
Abstrak:
Deteksi cacat merupakan langkah penting dalam kontrol kualitas tekstil dan pakaian jadi. Sistem deteksi cacat yang efisien dapat memastikan kualitas keseluruhan dari proses dan produk yang dapat diterima oleh konsumen. Teknik yang ada untuk deteksi cacat waktu nyata cenderung bervariasi sesuai dengan proses pembuatan yang unik, cacat fokus, dan algoritme komputasi. Meskipun kebutuhannya tinggi, penelitian terkait proses deteksi cacat kain cetak otomatis tidak lazim dalam literatur akademik. Penelitian ini mengusulkan metodologi baru yang menunjukkan penerapan jaringan saraf convolutional (CNN) untuk mengklasifikasikan cacat pencetakan berdasarkan gambar kain yang dikumpulkan dari industri. Penelitian ini juga mengintegrasikan jaringan pembelajaran mendalam visual geometric group (VGG), DenseNet, Inception dan Xception untuk membandingkan kinerja model. Hasilnya menunjukkan bahwa model berbasis VGG berkinerja lebih baik dibandingkan dengan model CNN sederhana, menunjukkan janji untuk deteksi cacat otomatis (ADD) dari kain cetak yang dapat meningkatkan profitabilitas dalam rantai pasokan mode.
https://doi.org/10.1080/17543266.2021.1925355
Moore, Marguerite, Parrillo-Chapman, Lisa, Chakraborty, Samit. (2022). Automatic defect detection for fabric printing using a deep convolutional neural network. (pp. 142-157)

-------------------------------------------------------------------------------------------------
Abstrak:
Manufaktur aditif memungkinkan pembuatan komponen dengan geometri kompleks, sehingga membuka ruang desain dari skala komponen hingga skala mikroarsitektur. Selain itu, dapat memperkirakan gradien sifat material sehubungan dengan perubahan struktur skala mikro membuat pengganti berbasis 3D-CNN yang diusulkan mudah beradaptasi dengan kerangka kerja pengoptimalan topologi multiskala yang ada. Melalui simulasi ekstensif, dengan membandingkan SIMP dan metode berbasis pengganti yang ada, kami menunjukkan keuntungan dari model pengganti berbasis 3D-CNN yang diusulkan.
https://doi.org/10.1080/17452759.2021.1913783
Guo Yilin, Jerry Fuh Ying Hsi & Lu Wen Feng. (2021). Multiscale topology optimisation with nonparametric microstructures using three-dimensional convolutional neural network (3D-CNN) models, Virtual and Physical Prototyping, (pp. 306-317)

Anonim mengatakan...

Nama : Nadya Adinda Majesty
NPM : 57419255
Kelas : 4IA16

Paper 1 :
Novia Farhan Nissa, Angelia Janiati, Nilam Cahya, Anton. Puji Astuti. Application of Deep Learning Using Convolutional Neural Network
(CNN) Method for Women’s Skin Classification. Vol. 8, No. 1, May 2021.
Abstract :
Purpose: Facial skin is the skin that protects the inner part of the face such as the eyes, nose, mouth, and other. The skin of the face consists of some type, among others, normal skin, oily skin, dry skin, and combination skin.
Methods: In this study, the method of the Convolutional Neural Network (CNN) is an appropriate method to classify the type of the skin of women of age 20 – 30 years by following a few stages using Python 3.5 with a depth of three layers.
Novelty: It can be concluded that the use of the method of CNN in automatic object recognition in distinguishing the type of leather as a material consideration in determining the object of the image.

Paper 2 :
Imania Ayu Anjani, Yulinda Rizky Pratiwi and Norfa Bagas Nurhuda S. Implementation of Deep Learning Using Convolutional Neural Network Algorithm for Classification Rose Flower. 2021.
Abstract :
Flora in Indonesia has about 25% of the species of flowering plant species present in the world. Roses are one type of flowering plants and are usually used as an ornamental plant
that has a thorny stem. Roses have more than 150 species. In Indonesia there are several flower
gardens that is larger than the others. One of the famous flower garden in Indonesia is located
on Malang city, East Java. The flower garden in Malang has several varieties of many roses and
has a large production of roses. To help the sales system of roses there, the researchers want to
create a program that can classify the type of roses in order to help simplify the system of
automatic sales of roses without through manual sorting. So that will accelerate the sale of roses
with an automated system. Ordinary people with limited botanical knowledge usually don’t
know how to classify the flowers just by looking at them. To classify the flowers properly, it is
important to provide enough information, and one of them is the name of it. Convolutional
Neural Network (CNN) is one method of deep learning that can be used for image classification
process. The CNN design is motivated by the discovery of the visual mechanism, the visual
cortex present in the brain. CNN has been widely used in many real-world applications, such as
Face Recognition, Image Classification and Recognition, and Object Detection, because this is
one of the most efficient method for extracting important features.

Paper 3 :
Sakshi Indolia, Anil Kumar Goswamib, S. P. Mishra, Pooja Asopa. Conceptual Understanding of Convolutional Neural Network- A Deep Learning Approach. 2018.
Abstract :
Deep learning has become an area of interest to the researchers in the past few years. Convolutional Neural Network (CNN) is a deep learning approach that is widely used for solving complex problems. It overcomes the limitations of traditional machine learning approaches. The motivation of this study is to provide the knowledge and understanding about various aspects of CNN. This study provides the conceptual understanding of CNN along with its three most common architectures, and learning algorithms.

[4IA16, Nadya Adinda Majesty, 57419255]