REV Journal on Electronics and Communications
Not a member yet
230 research outputs found
Sort by
Performance Analysis of Gradient Inversion Attack in Federated Learning with Healthcare Systems
Get PDFFederated learning (FL) is widely applied to healthcare systems with the primary aim of keeping the privacy of patient's data while improving classification quality by using knowledge from multiple participants. However, the training images are believed to be embedded into the shared gradient, which indicates a privacy risk when sharing the gradient with other participants in FL. Therefore, this work aims to design and evaluate an image recovery attack on medical images. More specifically, dummy images are trained to match the dummy gradient to the shared gradient while maintaining the smoothness and naturalness of reconstructed images. On the adversary side, an optimization problem is formulated with variables of dummy images and network parameters treated as constants. We evaluate the gradient attack on two medical datasets and reconstructed images clearly show the details of chest X-ray and MRI images including bone and blood vessels of captured areas. Our work aims to increase the awareness of people on sharing the gradient in FL, especially in healthcare systems
A New Framework for Cyber Risk Assessment for Industry 4.0 and Recommendations for Vietnam
Get PDFIndustry 4.0 has brought huge benefits to a wide range of industries. Its development, however, has raised more cyber security risks in both Information Technology (IT) and Operational Technology (OT) systems. In this paper, potential cyber vulnerabilities and threats in manufacturing in Industry 4.0 are briefly reviewed based on the architecture and operating principle of the manufacturing system in Industry 4.0. Criteria for cyber risk assessment for both IT and OT are reviewed via different standards. We then provide recommendations for cyber risk assessment and discuss a new framework for IT/OT risk assessment in Vietnam
Poliface: A Multi-pose Synchronous Imaging System
Get PDFTo enhance the accuracy of face recognition technology in real-world scenarios, it is necessary to train deep learning models on datasets that contain a large number of labeled human face images under multiple poses, lighting, and accessory variations. In this paper, we introduce a novel acquisition system named the Poliface. This system can capture multiple high-resolution images simultaneously around the human head. We designed this system with a well-built aluminum structure, control electronic circuits, and high-performing in-house software. The results demonstrate the precise operation and exceptional stability of this system. Using this Poliface system, we have collected over 6 million photos, which can be used to train and evaluate facial recognition models, and exploited for three-dimensional (3D) virtual face reconstruction
Performance Evaluation of a Decentralized Learning Architecture for PCB Defect Classification
Get PDFPrinted circuit board (PCB) defect detection, which is an important task in industrial factories, receives great attention from both researchers and practitioners. To achieve high detection accuracy, the traditional training method requires data collection from multiple industrial factories. However, in practice, factories possess their own data and do not want to share the private data with other participants. Therefore, we introduce a decentralized learning method that makes use the knowledge of clients in the system. By leveraging the federated learning technique, a consensus global detection model can be produced while maintaining data privacy. We have conducted extensive experiments to evaluate the detection performance under various learning methods: federated learning, centralized learning, and local learning. We also compare the detection performance of two well-known detection models: YOLOv5 and YOLOv8. The experimental results show that the federated learning based method yields better detection performance than the local learning
UIT-MLReceipts: A Multilingual Benchmark for Detecting and Recognizing Key Information in Receipts
Get PDFThe 4.0 industrial evolution has paved the way for development potential and revolution in Vietnam. In this movement, digitization appears to be necessary to transform numerous traditional economic sectors. It will provide valuable digital data for many automation applications and decision-making processes. Particularly in the retail industry, data has long played a vital factor. Hence, digitizing documents such as receipts can help businesses in management and enterprise development. Nevertheless, the digital transformation process is still slow because of the shortage of cleaned datasets for this type of document. This paper introduces a new dataset named UIT-MLReceipts for extracting key information in receipts. The task includes two sub-tasks: Receipt Text Detection (RTD) and Receipt Text Recognition (RTR). We thoroughly evaluate current state-of-the-art Receipt Text Detection using Faster R-CNN, YOLOv3, YOLOF, and Faster R-CNN with Precise RoI-Pooling on our dataset. To evaluate the performance of Receipt Text Recognition, we experiment with two text recognition baselines: RobustScanner and SATRN. Experimental results indicate that the Faster R-CNN with Precise RoI-Pooling outperforms the competitors and achieves the best mean Average Precision (mAP) score at 51.6% in the Receipt Text Detection task. With the Receipt Text Recognition task, results show that SATRN performs better.
Multi-Objective Optimization for IRS-Aided Multi-user MIMO SWIPT Systems
Get PDFIn this paper, we investigate an intelligent reflecting surface (IRS) assisted simultaneous wireless information and power transfer (SWIPT) system in which users equipped with multiple antennas exploit power-splitting (PS) strategies for simultaneously information decoding (ID) and energy harvesting (EH). Different from the majority of previous studies which focused on single objective optimization problems (SOOPs) and assumed the linearity of EH models, in this paper, we aim at studying the multi-objective optimization problem (MOOP) of the sum rate (SR) and the totalharvested energy (HE) subject to the maximum transmit power (TP) constraint, the user quality of service (QoS), and HE requirements at each user with taking a practical non-linear EH (NLEH) model into consideration. To investigate insightful tradeoffs between the achievable SR and total HE, we adopt the modified weighted Tchebycheff method to transform the MOOP into a SOOP. However, solving the SOOPs and modified SOOP is mathematically difficult due to the non-convexity of the object functions and the constraints of the coupled variables of the base station (BS) transmit precoding matrices (TPMs), the user PS ratios (PSRs), and the IRS phase shift matrix (PSM). To address these challenges, an alternating optimization (AO) framework is used to decompose the formulated design problem into sub-problems. In addition, we apply the majorization-minimization (MM) approach to transform the sub-problems into convex optimization ones. Finally, numerical simulation results are conducted to verify the tradeoffs between the SR and the total amount of HE. The numerical results also indicate that the considered system using the IRS with optimal phase shifts provides considerable performance improvement in terms of the achievable SR and HE as compared to the counterparts without using the IRS or with the IRS of fixed phase shifts
Decoder-ROI based Versatile Video Coding for Multi-Object Tracking Vision Task
Get PDFThe video encoding standards High Efficiency Video Coding (HEVC) and, more recently, Versatile Video Coding (VVC) have introduced significant advancements in multimedia communication applications, such as video conferencing, broadcasting, and notably, E-learning. However, recent developments in artificial intelligence (AI) and big data have given rise to an urgent need for a specialized video encoding model designed specifically for image and video analysis applications using machine vision. In this paper, we propose a novel video encoding approach that effectively combines the ROI Coding algorithm and the VVC encoding model. The proposed method identifies regions of interest within video frames through fundamental and deep features. Based on this, we propose an adaptive compression method for each frame block, ensuring both the execution performance of machine learning applications and minimal data encoding requirements. To achieve new coding scheme without adding bitrate, New feature extraction approach are utilizing only decoded information (Decoder-ROI). The results demonstrate that the Decoder-ROI achieved significant compression rate improvement when compared to standard and relevant VCM schemes. Furthermore, ROI exploitation contributes to a 3.25\% reduction in encoding time compared to the baseline VVC encoding standard.
Enhancing Feature Selection in MCI Diagnosis using FDG-PET Images: Leveraging Multiple Simple Autoencoder Architectures
Get PDFAlzheimer’s Disease (AD) is the most common type of neurodegenerative brain disease in elderly people. Early diagnosis of AD is crucial for providing suitable care. Positron Emission Tomography (PET) images and machine learning can be used to support this purpose. In this paper, we present a method for ranking the effectiveness of brain regions of interest (ROIs) to distinguish between stable mild cognitive impairment (sMCI) from progressive mild cognitive impairment (pMCI) in brain PET images based on AutoEncoder (AE). Experiments on the ADNI dataset show that our proposed method significantly improves classifier performance when compared to other popular feature ranking methods such as Fisher score, t-score, and LASSO. Our results suggest that instead of focusing on designing a complex AE structure, we can also use simple-but-multiple AEs for feature ranking. The proposed method could be easily applied to any image dataset where a feature selection is needed
Attention-Based BiGRU Model for Real-time Sign Language Translation Applications
Get PDFSign language applications provide an important key to solving communication problems for deaf community and normal hearing people. Current research problem usually focuses on improving communication access between deaf and hearing people. In this study, we consider real-time communication context from deaf to hearing people, and thus we propose an attention-based bidirectional gated recurrent unit (A-BiGRU) model which demonstrates on trading-off among an precision performance, and computational efficiency which includes training time, testing time, and system resources on extended the American Sign Language Gloss (E-ASLG-PC12) dataset. The results shown that our proposal has a significant performance improvement in term of training time, testing time, system resources, comparing to attentionbased bidirectional long-short term memory (A-BiLSTM), and the other moderns of sequence to sequence models. Moreover, precision performance of our proposal model achieve closer to that of the complex architecture, A-BiLSTM. Thus, we believe that our proposed model is a suitable and potential candidate for real-time translation applications as well as and lower computational devices when they solve the communication problems from deaf to normal hearing people direction. Keywords– computational efficiency (CE), attention-based bidirectional gated recurrent unit (A-BiGRU), sign language translation applications (SLTA)