|
|
|
Issues: Hệ thống và Thiết bị thông minh Vol 34.1 (01/2024)
|
1. An Efficient Face Recognition System Based on Edge Processing Using GPUs
Ha Xuan Nguyen*, Dong Nhu Hoang, Hung Trung Nguyen, Hai Minh Ngo, Tuan Minh Dang
In this work, an efficient and accurate face recognition system based on edge processing using GPUs was completely developed. A complete pipeline that contains a sequence of processing steps, including preprocessing, face feature extraction, and matching, is proposed. For processing steps, lightweight deep neural models were developed and optimized so that they could be computationally accelerated on an embedded hardware of Nvidia’s Jetson Nano. Besides the core processing pipeline, a database, as well as a user application server were also developed to fully meet the requirements of readily commercialized applications. The experimental evaluation results show that our system has a very high accuracy based on the BLUFR benchmark, with a precision of 98.642%. Also, the system is very computationally efficient, as the computing time to recognize an ID in a dataset of 1171IDs with 10141 images on the Jetson Nano is only 165ms. For the critical case, the system can process 4 camera streams and simultaneously recognize a maximum of 40 IDs within a computing time of 458ms for each ID. With its high-speed and accuracy characteristics, the developed system has a high potential for practical applications.
Article Code:23012
# Face Recognition, Deep Learning, GPUs, Edge Processing
Page: 1-8
Field: Information and Communication Technology
|
2. Solving Resource Allocation Problem in Wifi Network by Dantzig-Wolfe Decomposition Algorithm
Ta Anh Son*, Nguyen Thi Thuy
Online advertising and advancements is a recent trend in marketing technology, in this context we consider a new form of contract which allows advertisers to specify in the wifi system. Based on the structure of the system, we have to organize and manage resource allocation such that the guaranteed display is satisfied. We introduce a new mathematical model and develop an optimization framework that aims to optimize “fairness” of allocation each campaign over its targeted location. Because of large scale problem, the Dantzig-Wolfe decomposition is proposed for solving it. Dantzig-Wolfe decomposition is a technique for dealing with large scale linear programming and modified to solve linear integer programming, nonlinear programming. Especially, it is used mostly in linear programming when its size is very large, and its structure is appropriate. The technique has been successfully applied in a variety of contexts. In this paper, we introduce a new model of a resource allocation problem in Wifi network and represent Dantzig-Wolfe decomposition for solving this problem by dividing the number of advertisement impressions when users access the wifi network. The numerical simulation shows the efficiency of our proposed method.
Article Code:23013
# Dantzig-Wolfe Decomposition, Resource Allocation, Wifi Network, Online Advertising
Page: 9-15
Field: Applied Mathematics and Informatics
|
3. Credit Card Service Churn Prediction by Machine Learning Models
Tran Hoang Hai*, Vu Van Thieu, Doan Minh Hieu
This paper presents a study on the application of basic machine learning models for churn customer classification. Churn prediction is an essential task in customer retention for businesses, and accurate identification of customers who are likely to churn can significantly impact the organization's revenue and customer satisfaction. In this study, we explore the performance of various machine learning models, including K-Nearest Neighbor, Random Forest, Adaboost and a deep learning model which is CNN-1D. We use the BankChurners dataset, then we predict the probability that customers abandoning bank services such as credit card services. We evaluate the models basing on various performance metrics such as accuracy, precision, recall, and F1-score. The result demonstrates the potential of basic machine learning models for churn customer classification and provides insights into the key factors contributing to customer churn.
Article Code:23042
# customer churn, machine learning, classification, credit card
Page: 16-22
Field: Information and Communication Technology
|
4. EDDS-Unet: An Encoder-Decoder Double Skip Connection Scheme for Skin Lesion Segmentation
Thi-Thao Tran, Minh-Nhat Trinh, Nhu-Toan Nguyen, Van-Truong Pham*
This paper presents an approach for effective skin lesion segmentation from dermatoscopic images. Aiming at transferring the weights trained from a network originally designed for image classification task, this study proposes to utilize the first layers of EfficientNet as the encoding layers of a U-Net based architecture. Besides, we introduce an encoder-decoder double skip connection scheme, a new skip connection architecture for extracting useful spatial details of skin lesions from the encoding layers. By the double skip scheme, the approach not only fuses information from the encounter layer in the encoder path to the corresponding layer in the decoder path, but also takes into account information of the proceeding encoding layer. In addition, we propose a new decoder network using the Residual blocks and Convolutional Block Attention Module (CBAM) blocks to handle the gradient vanishing problem as well as penalize the weight of each layer. The proposed Encoder-Decoder Double Skip with the Unet architecture, namely EDDS-Unet, has shown promising performance when evaluated on the official ISIC 2017 challenge and the PH2 databases. The proposed method achieves high evaluation scores with the Dice Similarity Coefficients of 0.907 for the ISIC 2017 and 0.950 for the PH2 databases without pre-or post-processing steps.
Article Code:23060
# skin lesion segmentation, skip connection, u-net, efficientNet
Page: 23-30
Field:
Electrical Engineering
|
5. Study of the Dynamics of Automated Guided Vehicle Using Mecanum Omnidirectional Wheel
Tong Xuan Loc, Truong Dang Viet Thang, Ho Huu Hai*
The paper presents the result of the dynamics simulation of automated guided vehicles (AGVs) using four Mecanum omnidirectional wheels when the tilt angle of each wheel’s roller is changed. Mecanum omnidirectional wheels have the ability to work flexibly, but operators are generally concerned about the vehicle’s efficiency issues. The performance of the wheel significantly influences the extensive application range of the vehicle. The method of modeling automated guided vehicles and Mecanum wheels is applied in order to build its kinematic model and dynamic model with omnidirectional wheels when the roller’s angle of incline is changed. Based on Lagrange’s equation type II, the equation of motion of the vehicle is established in accordance with the actual situation that the vehicle is being applied. The rolling friction coefficient and the adhesive coefficient of Mecanum wheels were measured experimentally on different road surfaces, including concrete, wood, brick, and stone. The study's findings successfully address efficiency challenges in vehicles using omnidirectional wheels when varying the incline angle of each roller and applying fuzzy logic for trajectory control in autonomous vehicles.
Article Code:23064
# Automated guided vehicles (AGVs), Fuzzy logic control, Lagrange’s equation type II, Mecanum omnidirectional wheel.
Page: 31-41
Field: Mechanical Engineering
|
6. Fuzzy observer-based control design for Rotary Inverted Pendulum Using Takagi-Sugeno Model
Nguyen Thi Van Anh, Dong Bao Trung, Phan Bao Ngoc, Quy-Thinh Dao*
This paper introduces fuzzy observer-based control for rotary inverted pendulum systems, renowned for their inherent instability and complexity. We leverage the Takagi-Sugeno (T-S) fuzzy model, which involves the incorporation of local linear models described by fuzzy rules, thus enabling precise and stable control. The Takagi-Sugeno (T-S) fuzzy model, a versatile framework renowned for its suitability in complex control systems, is central to our approach. The significance of observers in accurately estimating unmeasurable states is underlined, with a focus on elucidating the theoretical foundations of fuzzy observers and their role in bolstering control robustness. Additionally, we introduce the integration of Linear Matrix Inequalities (LMIs) and Parallel Distributed Compensation (PDC) for efficient determination of observer and control gains. These advanced tools work in tandem to empower T-S observer control, ensuring both precision and robustness. This paper shows the potential of fuzzy observer-based control and achieving stability and high-performance control of rotary inverted pendulum systems. The effectiveness of the proposed method is validated through simulation results.
Article Code:23068
# Takagi-Sugeno fuzzy model, observer control, linear matrix inequality, rotary inverted pendulum
Page: 42-50
Field:
Electrical Engineering
|
7. Control of the Tower Crane using Input Shaping-Sliding Mode Control
Minh Duc Duong*, Quy Thinh Dao, Minh Dung Le, Duy Long Le
Tower cranes are widely used for moving heavy goods, materials, or tools around a site. They help to speed up construction, save time and manpower in a process. However, a significant problem of tower cranes is oscillatory behavior, which can adversely impact safety and delivery accuracy. This paper proposes sliding mode control (SMC) combined with Input Shaping (IS) for controlling tower cranes. Sliding mode control itself can be used to control a tower crane to obtain position precision and vibration suppression. However, the selection of controller parameters may be difficult and the required control effort is high. In addition, chattering may occur. With the combination of input shaping, these problems can be overcome. The control parameter range is extended and the required control effort is reduced when input shaping is applied. In addition, input shaping also helps to reduce load vibration and chattering. Simulations in Matlab-Simulink have been done and the simulation results show the effectiveness of the proposed control algorithm.
Article Code:23002
# Tower Crane, Sliding mode control, Input Shaping, Vibration Suppression Control
Page: 51-57
Field:
Electrical Engineering
|
8. An Agent-Based Model for Simulating the Interactions of Tigers, Leopards, and Wild Boars to Support Wildlife Conservation Planning
Phung Anh Hung, Nguyen Duc Hung, Nguyen Phuong Thùy*
Wildlife conservation is a pressing global concern, with the need to create and manage protected areas where multiple species can coexist without facing the threat of extinction. In this paper, we proposed an agent-based model that simulates the interactions and life activities of tigers, leopards, and wild boars within a 400 km2 area, approximately the area of standard conservation. The model incorporates the three animal species' physical characteristics and behavioral traits to analyze their mutual influence within the environment. The emergence results indicate that changes in the wild boar population size affect the survival of tigers and leopards, with population increases or decreases in one species impacting the others. Moreover, when tigers or leopards become overly dominant in population size, they consume more wild boar, leading to increased competition and potential extinction of the other species. Additionally, the study highlights the importance of the non-uniform distribution of plant food resources in conservation areas, emphasizing that wild boar food resources should occupy at least 70% of the site. These findings are valuable for understanding ecological dynamics, informing conservation area design, and predicting scenarios requiring human intervention to maintain species balance.
Article Code:23047
# agent-based modeling, ODD protocol, tiger-leopard-wild boar system, foodweb
Page: 58-65
Field: Applied Mathematics and Informatics
|
 | Loading… |
|
|
|
|