检索结果-内蒙古大学图书馆

10th International Conference on Digital Arts, Media and Technology and 8th ECTI Northern Section Conference on Electrical, Electronics, computer and Telecommunications engineering, ECTI DAMT and NCON 2025

作者： Mekruksavanich, Sakorn Hnoohom, Narit Phaphan, Wikanda Jitpattanakul, Anuchit University of Phayao School of Information and Communication Technology Department of Computer Engineering Phayao Thailand Mahidol University Image Information and Intelligence Laboratory Department of Computer Engineering Faculty of Engineering Nakhon Pathom Thailand King Mongkut's University of Technology North Bangkok Research Group in Statistical Learning and Inference Department of Applied Statistics Faculty of Applied Science Bangkok Thailand King Mongkut's University of Technology North Bangkok Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Faculty of Applied Science Bangkok Thailand

ISBN: (纸本)9798331543273

With the rise of wearable sensors and smart devices, human activity recognition (HAR) has become a vital research area in ubiquitous computing. Although many studies report high accuracy using k-fold cross-validation, these results often do not reflect actual generalization interpretation due to subject-dependent data leakage, where models test on activities from the subjects they were trained. This study compares traditional k-fold cross-validation with leave-one-subject-out (LOSO) validation using the WISDM dataset, highlighting the importance of proper validation techniques in HAR systems. We implemented and evaluated five advanced deep learning models - convolutional neural network (CNN), long short-term memory (LSTM), bidirectional LSTM (BiLSTM), gated recurrent unit (GRU), and bidirectional GRU (BiGRU) - on windowed raw sensor data. Our experiments showed significant differences between validation methods. The BiGRU model achieved 97.91% accuracy with k-fold cross-validation and 98.02% with LOSO validation, while the CNN model achieved only 92.55% and 93.44%, respectively. These results underscore the impact of both model architecture and validation approach on performance. Our findings stress the need for subject-independent validation strategies like LOSO to develop truly generalizable HAR systems. © 2025 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Image Classification and Segmentation for Estimating Calories of Food 10

Image Classification and Segmentation for Estimating Calorie...

引用

作者： Chanthanuwatkun, Udomkan Jitpattanakul, Anuchit Hnoohom, Narit Faculty of Engineering Mahidol University Image Information and Intelligence Laboratory Department of Computer Engineering Nakhon Pathom Thailand Faculty of Applied Science King Mongkut's University of Technology Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics North Bangkok Thailand

ISBN: (纸本)9798331543273

Accurate estimation of calorie content in meals is essential for promoting healthy dietary practices and supporting healthcare efforts. However, the lack of nutritional knowledge poses challenges for individuals in estimating the calorie content of foods. This study aims to develop a deep learning-based model for estimating the calorie content of real foods through image classification and segmentation techniques. Image classification was employed to identify various one-dish Thai menus, while image segmentation categorized food components into fats, carbohydrates, and proteins. A new dataset, consisting of 15 one-dish Thai menus, was created using a combination of publicly available datasets (THFOOD-50 and FoodyDudy) and self-collected images. Nutritional analysis was based on the Thai nutritional table from the Bureau of Nutrition, Department of Health, Ministry of Public Health. We compared the performance of YOLOv5x, YOLOv8x, and YOLOv9e on the ThaiFood dataset using identical hyperparameters. The results showed that YOLOv8x achieved the highest mAP@0.5 of 0.883, with an accuracy of 0.829 and a recall of 0.902. YOLOv5x closely followed with a mAP@0.5 of 0.881, an accuracy of 0.822, and a recall of 0.945. YOLOv9e demonstrated competitive performance, achieving a mAP@0.5 of 0.846, an accuracy of 0.775, and a recall of 0.91. These findings highlight YOLOv8x as the most suitable model for Thai food classification and detection tasks under the evaluated conditions. © 2025 IEEE.

关键词： Nutrition

来源：评论

学校读者我要写书评

暂无评论

Deep Reinforcement Learning of Robotic Manipulation for Whip Targeting

Deep Reinforcement Learning of Robotic Manipulation for Whip...

引用

2025 IEEE International Conference on Simulation, Modeling, and Programming for Autonomous Robots, SIMPAR 2025

作者： Bai, Xiang Wang, Junyi Xiong, Xiaofeng Boukas, Evangelos Technical University of Denmark Applied Mathematics and Computer Science Copenhagen Denmark Technical University of Denmark Automation and Control Group Electrical and Photonics Engineering Copenhagen Denmark University of Southern Denmark Sdu Biorobotics Mærsk Mc-Kinney Moller Institute Odense Denmark

ISBN: (纸本)9798331516857

关键词： Robotic arms

来源：评论

学校读者我要写书评

暂无评论

Privacy Vulnerabilities in Marginals-based Synthetic Data

Privacy Vulnerabilities in Marginals-based Synthetic Data

引用

Secure and Trustworthy Machine Learning (SaTML), IEEE Conference on

作者： Steven Golob Sikha Pentyala Anuar Maratkhan Martine De Cock School of Engineering and Technology University of Washington Tacoma USA Department of Applied Mathematics Computer Science and Statistics Ghent University Belgium

ISBN: (数字)9798331517113

ISBN: (纸本)9798331517120

When acting as a privacy-enhancing technology, synthetic data generation (SDG) aims to maintain a resemblance to the real data while excluding personally-identifiable information. Many SDG algorithms provide robust differential privacy (DP) guarantees to this end. However, we show that the strongest class of SDG algorithms-those that preserve marginal probabilities, or similar statistics, from the underlying data-leak information about individuals that can be recovered more efficiently than previously understood. We demonstrate this by presenting a novel membership inference attack, MAMA-MIA, and evaluate it against three seminal DP SDG algorithms: MST, PrivBayes, and Private-GSD. MAMA-MIA leverages knowledge of which SDG algorithm was used, allowing it to learn information about the hidden data more accurately, and orders-of-magnitude faster, than other leading attacks. We use MAMA-MIA to lend insight into existing SDG vulnerabilities. Our approach went on to win the first SNAKE (SaNitization Algorithm under attacK … ε) competition.

关键词： Differential privacy Machine learning algorithms Technology Machine learning Probability Inference algorithms Synthetic data Identification of persons

来源：评论

学校读者我要写书评

暂无评论

A Deep Multi-Task Learning Network for Activity Recognition and User Identification Using Smartphone Sensors 16

A Deep Multi-Task Learning Network for Activity Recognition ...

引用

16th International Conference on ENTERprise Information Systems, CENTERIS 2024 - 12th International Conference on Project MANagement, ProjMAN 2024 - 14th International Conference on Health and Social Care Information Systems and Technologies, HCist 2024

作者： Mekruksavanich, Sakorn Phaphan, Wikanda Jitpattanakul, Anuchit Department of Computer Engineering School of Information and Communication Technology University of Phayao Phayao56000 Thailand Research Group in Statistical Learning and Inference Department of Applied Statistics Faculty of Applied Science King Mongkut's University of Technology North Bangkok Bangkok10800 Thailand Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Faculty of Applied Science King Mongkut's University of Technology North Bangkok Bangkok10800 Thailand

Human activity recognition (HAR) and user identification utilizing smartphone sensors are crucial in domains such as wellness tracking, tailored services, and security. Conventional approaches handle these activities as distinct entities, leading to less-Than-ideal performance and restricted applicability. This research presents a novel deep multi-Task learning network that simultaneously performs activity recognition and user identification using smartphone sensor data. By leveraging shared representations and task-specific information, the network enhances accuracy. It combines a shared convolutional neural network (CNN) with bidirectional long short-Term memory (BiLSTM) layers to extract features from raw sensor data. The CNN identifies patterns, while the BiLSTM captures sequential dependencies and contextual information. These features are processed by fully connected layers for both activity recognition and user identification. A multi-Task loss function is proposed to optimize both tasks based on their complexity and importance. The approach is tested on the UCI-HAR dataset, containing diverse activities from multiple users. Experimental results show that the proposed CNN-BiLSTM model achieves 97.64% accuracy in activity recognition and 82.59% in user identification, surpassing current state-of-The-Art methods. This demonstrates the effectiveness of the multi-Task learning framework and the advantage of using the shared CNN-BiLSTM architecture for improving both tasks. © 2025 The Author(s).

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

LowResIR-Net: Lightweight Residual Network for ADL Recognition Using Low-Resolution Infrared Sensor Data 16

LowResIR-Net: Lightweight Residual Network for ADL Recogniti...

引用

Detecting activities of daily living (ADL) is crucial for supported living and medical monitoring. Traditional approaches rely on high-resolution sensors and computationally intensive algorithms, limiting their scalability and practical use in real-world settings. This research introduces LowResIR-Net, a compact deep residual network architecture specifically tailored for recognizing ADL utilizing low-resolution infrared sensor data. LowResIR-Net efficiently extracts key features from low-dimensional data by leveraging the advantages of infrared sensing, such as privacy protection and light resilience. Its architecture combines residual connections with depth-wise convolutions, enhancing feature learning while maintaining a compact model size. The model’s performance is evaluated using the Coventry-2018 dataset, containing low-resolution infrared data from three residential sensor locations. Our tests have shown that in all three cases, LowResIR-Net consistently achieves superior performance compared to other cutting-edge deep learning models, such as CNN, LSTM, BiLSTM, GRU, and BiGRU. LowResIR-Net attains the most excellent accuracies of 90.18%, 93.75%, and 93.30%, and F1-scores of 81.81%, 89.55%, and 88.92% for scenarios I, II, and III, respectively. Furthermore, it achieves these results with fewer parameters than most other models. LowResIR-Net’s strong performance highlights its ability to accurately capture complex patterns in low-resolution infrared data for ADL detection. This study demonstrates the potential of lightweight deep learning and low-resolution infrared sensors in developing privacy-focused and computationally efficient wellness monitoring systems. © 2025 The Author(s).

关键词： Infrared detectors

来源：评论

学校读者我要写书评

暂无评论

Representation with Minimized Max-Error in Optimal Piecewise Linear Approximation of Time Series Data 25th

Representation with Minimized Max-Error in Optimal Piecewi...

引用

25th International Conference on Web Information Systems engineering, WISE 2024

作者： Zhao, Huanyu Li, Tongliang Wen, Shiting Shu, Zhenyu Deng, Ke Yang, Jian Pang, Chaoyi School of Computer Science and Technology Donghua University Shanghai China Institute of Applied Mathematics Hebei Academy of Sciences Shijiazhuang China Hebei Authentication Technology Engineering Research Center Shijiazhuang China Institute of Biology Hebei Academy of Sciences Shijiazhuang China School of Computer and Data Engineering NingboTech University Ningbo China RMIT University Melbourne Australia Macquarie University Sydney Australia

ISBN: (纸本)9789819605781

In the past two decades, Piecewise Linear Approximation under maximum error (max-error) bound (PLA∞) has been intensively studied for effective qualified representation and analysis of time series data. It divides a time series into fragments and then represents each fragment with a straight line to approximate the data points of that time slot. In this paper, to elevate the representation quality in the optimal PLA∞ results, we present a linear-time algorithm FindMin to construct the unique line representative of minimized maximum error (min-max error) for each fragment. Through extended experimental tests, we demonstrate that the proposed algorithm is very efficient in execution and achieves better performances than the state-of-the-art solutions. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Time series analysis

来源：评论

学校读者我要写书评

暂无评论

Enhancing Performances of Deep Neural Networks with Ensemble Learning Methods for Complex Human Activity Recognition Using Wearable Sensors 10

Enhancing Performances of Deep Neural Networks with Ensemble...

引用

作者： Mekruksavanich, Sakorn Jitpattanakul, Anuchit University of Phayao School of Information and Communication Technology Department of Computer Engineering Phayao Thailand King Mongkut's University of Technology North Bangkok Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Faculty of Applied Science Bangkok Thailand

ISBN: (纸本)9798331543273

Sensor-based human activity recognition (HAR) is a widely explored research area focused on identifying human behaviors by analyzing sensor-generated data, particularly one-dimensional time series signals. Such HAR systems have many real-world applications, including elderly health monitoring and athlete performance evaluation. This paper presents Ens-DeepNet, a novel ensemble learning approach designed for recognizing complex human activities using data from wearable sensors. While deep neural networks have proven effective for HAR tasks, their performance can be further improved through ensemble strategies. To this end, we propose an ensemble architecture that combines five distinct deep learning models - convolutional neural network (CNN), long short-term memory (LSTM), bidirectional LSTM (BiLSTM), gated recurrent unit (GRU), and bidirectional GRU (BiGRU). These models are integrated using a majority voting scheme to boost classification accuracy. The proposed method is evaluated using the UTwente dataset, which includes sensor data collected from ten individuals via tri-axial accelerometers and gyroscopes. Experimental findings demonstrate that the ensemble model achieves an impressive accuracy of 99.28%, surpassing the performance of each standalone model, whose accuracies range from 92.77%to 97.34%. This performance gain is particularly notable in recognizing complex activities such as eating, drinking coffee, smoking, and conversing. © 2025 IEEE.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

CAPE: Commitment-based Privacy-Preserving Payment Channel Scheme in Blockchain

引用

IEEE Transactions on Dependable and Secure Computing 2025年

作者： Gai, Keke Guo, Yunwei Yu, Jing Chan, Weilin Zhu, Liehuang Zhang, Yinqian Meng, Weizhi The School of Cyberspace Science and Technology Beijing Institute of Technology Beijing100081 China The Institute of Information Engineering Chinese Academy of Sciences Beijing China The Department of Computer Science and Engineering Southern University of Science and Technology Guangdong Shenzhen China Department of Applied Mathematics and Computer Science Cybersecurity Engineering Technical University of Denmark Kongens Lyngby Denmark

Ensuring scalability in cryptocurrency systems is significant in guaranteeing real-world utility along with the remarkable increment of cryptographic currency. As an alternative in solving scalability issue, payment channel allows users to deliver extensive offline transactions without uploading massive transaction details to the blockchain, such that increasing efficiency can be achieved. However, the implementation of payment channel still encounters privacy concerns when considering the publicly available transaction amounts and the potentials in mining associations between transaction parties. In this paper, we propose a novel payment channel scheme, entitled Commitment-based Anonymous Payment ChannEl (CAPE), to facilitate unlimited off-chain bidirectional payments while guaranteeing participants’ privacy. The proposed scheme adopts zero-knowledge proof (zk-SNARKs) and verifiable timed (VTD) commitments to ensure the anonymity of the relationship between on-chain and off-chain transactions, privacy of transaction amounts, and security of balances. We comprehensively formalize security definitions and present rigorous proofs for each security attribute. Experiment results further demonstrate the practical viability of CAPE. © 2025 IEEE. All rights reserved.

关键词： Differential privacy

来源：评论

学校读者我要写书评

暂无评论

A Comparative Analysis of IoT Data Using Machine Learning Algorithms 15th

A Comparative Analysis of IoT Data Using Machine Learning Al...

引用

15th International Conference on Soft Computing and Pattern Recognition, SoCPaR 2023 and 14th World Congress on Nature and Biologically Inspired Computing, NaBIC 2023

作者： Nagashree, B.A. Satheesh Kumar, S. Muthukumaran, V. Munrathnam, Meram Angeline Kavitha, M. Department of Computer Science and Applications School of CSA REVA University Bangalore India Department of Computer Science School of Applied Sciences REVA University Bangalore India Department of Mathematics College of Engineering and Technology SRM Institute of Science and Technology Chennai India Department of Mathematics Rajiv Gandhi University of Knowledge Technologies R.K. Valley Kadapa District Andhra Pradesh Idupulapaya 516330 India Department of Mathematics Voorhees College Vellore 632001 India

ISBN: (纸本)9783031889912

IoT (Internet of Things) is revolutionizing the commercial environment by acquiring vast amounts of data from various sources. However, processing and analyzing this data is challenging due to the large volume of IoT devices. AI-powered IoT creates intelligent machines that mimic human intelligence, aiding decision-making with minimal human input. Both individuals and professionals’ benefit from this combination, as IoT is concerned with communicating online, but AI allows machines to learn from data and experience. This study demonstrates the analysis of IoT data using machine learning algorithms and compares the accuracy of various machine learning algorithms used. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： AI data internet IoT Machine Learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：