检索结果-内蒙古大学图书馆

5th Research, Invention, and Innovation Congress: Innovative Electricals and Electronics, RI2C 2024

作者： Mekruksavanich, Sakorn Phaphan, Wikanda 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 Faculty of Applied Science Research Group in Statistical Learning and Inference Department of Applied Statistics Bangkok Thailand King Mongkut's University of Technology North Bangkok Faculty of Applied Science Intelligent and Nonlinear Dynamic Innovations Research Center Department of Mathematics Bangkok Thailand

ISBN: (纸本)9798331505998

Biometric identity verification is critical in security, healthcare, and personalized services. Photoplethysmography (PPG) signals, measuring blood volume changes in small vessels, offer a promising, non-invasive biometric method but often need more accuracy and robustness. To address this, we propose the ResNet-BiGRU-SE framework, which enhances PPG-based biometric identification through a s queeze-and-excitation (SE) attention mechanism. This architecture integrates convolutional neural networks (CNNs), bidirectional gated recurrent units (BiGRUs), and SE attention modules to extract spatial and temporal features from PPG signals. ResNet architecture in CNN layers extracts local patterns and mitigates vanishing gradients with residual connections, while BiGRU layers capture temporal dependencies and contextual information. The SE attention mechanism enhances focus on informative PPG signal parts by adjusting feature responses per channel. Evaluated on a standardized dataset of PPG signals from various patients, the ResNet-BiGRU-SE model demonstrates superior accuracy (95.26%), low loss (0.24), and a high F1-score (95.25 %), outper-forming recent methods. Comprehensive ablation studies confirm the model's robust and reliable performance. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Decentralized Federated Learning Loop with Constrained Trust Mechanism 1

引用

22nd International Conference on Artificial Intelligence and Soft Computing, ICAISC 2023

作者： Polap, Dawid Prokop, Katarzyna Srivastava, Gautam Chun-Wei Lin, Jerry Faculty of Applied Mathematics Silesian University of Technology Kaszubska 23 Gliwice44-100 Poland Department of Mathematics and Computer Science Brandon University BrandonR7A 6A9 Canada Western Norway University of Applied Sciences Bergen Norway

ISBN: (数字)9783031425059

ISBN: (纸本)9783031425042

Federated learning has made it possible to introduce parallel training of deep neural networks by multiple users. The use of model aggregation contributes to the generalization of it, although there is a possibility of attacks. An example of this is dataset poisoning. Hence, in this research paper, we propose the introduction of a constrained trust mechanism for individual clients. In addition, a decentralized approach makes it possible to increase the effectiveness of the training process by removing the server and reducing the risk of an attack on the transmitting model. The proposed modification of federated learning was subjected to performance tests and compared with other known solutions. The obtained results indicate an increase in safety and accuracy. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Detecting Face-Touching Gestures with Smartwatches and Deep Learning Networks 47

Detecting Face-Touching Gestures with Smartwatches and Deep ...

引用

47th International Conference on Telecommunications and Signal Processing, TSP 2024

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

ISBN: (纸本)9798350365597

Everyday actions like scratching one's nose or resting the chin on one's hand may facilitate the spread of germs and diseases. Detecting these gestures holds the potential for innovative health monitoring and disease prevention applications. However, the identification of face-touching poses challenges due to the variety of human gestures and the limitations in accuracy associated with wearable sensors. This study introduces an original deep-learning system to identify face-touching gestures using standard smartwatches' inertial measurement unit sensors. The system on the smartwatch captures and pre-processes multi-channel time-series data from the accelerometer to generate robust input features. We utilize a benchmark, the Face Touching dataset, to evaluate the recognition performance of deep learning networks, including our proposed network. Our approach proposes a hybrid deep residual network architecture tailored for sequence-based gesture classification using signals from the smartwatch sensors. In our experiments, the developed deep learning framework achieves an impressive F1-score of 97.53% in detecting face-touch gestures. The suggested system takes a step forward in advancing the practical applications of face-touch gesture detection to smartwatch-based health sensing and disease prevention. © 2024 IEEE.

关键词： Network architecture

来源：评论

学校读者我要写书评

暂无评论

Enhancing Sensor-based Human Activity Recognition Using Hybrid Deep Learning and Data Augmentation 5

Enhancing Sensor-based Human Activity Recognition Using Hybr...

引用

5th Research, Invention, and Innovation Congress: Innovative Electricals and Electronics, RI2C 2024

ISBN: (纸本)9798331505998

Human Activity Recognition (HAR) is essential in various applications, including wellness tracking, automated residences, and fitness monitoring. In the past few decades, sensor-based HAR has become increasingly popular due to advancements in technologies for sensors. Nevertheless, HAR networks' effectiveness dramatically depends on the caliber and volume of the training data, which is frequently restricted and unevenly distributed. This research introduces a novel deep learning method called Multihead-CNN-BiGRU, which integrates one-dimensional convolutional neural networks with bidirectional gated recurrent units (BiGRU) to improve the accuracy of sensor-based HAR. To tackle the problem of insufficient a nd uneven training data, we utilize the synthetic minority over-sampling technique (SMOTE) to augment the data. The suggested model is assessed using the publicly accessible WISDM dataset, which comprises sensor data from diverse human actions. The ID-CNN is employed for extracting localized characteristics from the sensor data, while the BiGRU gathers temporal dependencies and contextual information. The hybrid architecture allows the model to acquire spatial and temporal patterns efficiently. In addition, the SMOTE technique is utilized to create artificial samples of the underrepresented classes, thus equalizing the distribution of classes and enhancing the model's capacity to generalize. The experimental findings show that our hybrid strategy provides exceptional outcomes when paired with SMOTE data augmentation. It obtains the most excellent accuracy of 99.51 % and the highest F1-score of 99.49% compared to the most advanced approaches. The suggested framework provides a reliable and precise solution for sensor-based HAR, which opens up opportunities for improved applications in other fields. © 2024 IEEE.

关键词： Spatio-temporal data

来源：评论

学校读者我要写书评

暂无评论

A Sensor-Based Deep Learning Approach for Recognizing Daily and Work Activities in Open Environments for Sanitation Workers 9

A Sensor-Based Deep Learning Approach for Recognizing Daily ...

引用

Joint 9th International Conference on Digital Arts, Media and Technology with 7th ECTI Northern Section Conference on Electrical, Electronics, computer and Telecommunications Engineering, ECTI DAMT and NCON 2024

作者： Mekruksavanich, Sakorn Jantawong, Ponnipa Phaphan, Wikanda 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 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)9798350318241

Sanitation workers play a crucial role in maintaining public hygiene and cleanliness. Understanding their daily routines and job responsibilities is essential for improving their work environment, task distribution, and overall supervision. However, tracking sanitation workers in open environments remains a persistent challenge. This paper introduces a deep learning model named CNN-BiGRU-CBAM, designed to recognize sanitation workers' daily and working activities using smartwatch sensors. The model leverages advanced convolutional neural networks to extract highly informative features from multi-modal sensor data, including acceleration information, collected from smartwatches. The dataset used in this study is carefully curated and collected from sanitation workers during their daily work routines, encompassing activities such as walking, running, sweeping, lifting, and driving vehicles. These activities are systematically categorized to cover both work-related tasks and non-work activities. Experimental results demonstrate that the proposed model consistently achieves an average F1-score of over 94.50% across seven activity groups, outperforming baseline deep learning methods by an impressive margin of 5.62%. The model excels in accurately identifying operations in diverse and dynamic work settings. This research highlights the potential of smartwatches and deep learning in continuously analyzing sanitation workers' duties and developing improved working regulations that consider their well-being and daily tasks. © 2024 IEEE.

关键词： Sanitation

来源：评论

学校读者我要写书评

暂无评论

Harnessing Deep Learning for Activity Recognition in Seniors' Daily Routines with Wearable Sensors 47

Harnessing Deep Learning for Activity Recognition in Seniors...

引用

47th International Conference on Telecommunications and Signal Processing, TSP 2024

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

ISBN: (纸本)9798350365597

Wearable sensors are increasingly popular for monitoring the daily activities of older individuals, sparking significant interest in human activity recognition (HAR). Accurate and timely identification of these activities is crucial for providing personalized care and timely interventions to maintain the well-being of elderly individuals. However, existing HAR methods often need support with complex activity patterns, limited labeled data, and the need for real-time performance. To address these challenges, this study proposes a novel deep-learning approach called ResBiGRU for precisely identifying daily activities in older adults using wearable sensors. The ResBiGRU architecture combines residual connections with BiGRU networks to capture short-term and long-term dependencies in sensor data effectively. We also propose a data augmentation strategy to improve generalization performance. Evaluation of the HAR70+ dataset, consisting of activities performed by 18 elderly individuals, demonstrates superior performance of our ResBiGRU model with an accuracy of 97.39% and an F1-score of 97.74%. Furthermore, our model supports real-time inference, making it suitable for practical applications. Implementing the ResBiGRU approach could significantly enhance monitoring and care for older adults by accurately identifying their daily activities. This research advances HAR methods tailored for elderly individuals, opening new avenues for personalized healthcare and assisted living technologies. © 2024 IEEE.

关键词： Wearable sensors

来源：评论

学校读者我要写书评

暂无评论

Continuous Partitioning for Graph-Based Semi-Supervised Learning 38

Continuous Partitioning for Graph-Based Semi-Supervised Lear...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Holtz, Chester Chen, Pengwen Wan, Zhengchao Cheng, Chung-Kuan Mishne, Gal Halicioǧlu Data Science Institute University of California La Jolla San DiegoCA United States Department of Applied Mathematics National Chung Hsing University South District Taichung Taiwan Department of Mathematics University of Missouri ColumbiaMO United States Department of Computer Science University of California La Jolla San DiegoCA United States

Laplace learning algorithms for graph-based semi-supervised learning have been shown to suffer from degeneracy at low label rates and in imbalanced class regimes. Here, we propose CutSSL: a framework for graph-based semi-supervised learning based on continuous nonconvex quadratic programming, which provably obtains integer solutions. Our framework is naturally motivated by an exact quadratic relaxation of a cardinality-constrained minimum-cut graph partitioning problem. Furthermore, we show our formulation is related to an optimization problem whose approximate solution is the mean-shifted Laplace learning heuristic, thus providing new insight into the performance of this heuristic. We demonstrate that CutSSL significantly surpasses the current state-of-the-art on k-nearest neighbor graphs and large real-world graph benchmarks across a variety of label rates, class imbalance, and label imbalance regimes. Our implementation is available on github. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Unveiling the Power of Sparse Neural Networks for Feature Selection

arXiv

引用

arXiv 2024年

作者： Atashgahi, Zahra Liu, Tennison Pechenizkiy, Mykola Veldhuis, Raymond Mocanu, Decebal Constantin van der Schaar, Mihaela Faculty of Electrical Engineering Mathematics and Computer Science University of Twente Netherlands Department of Applied Mathematics and Theoretical Physics University of Cambridge United Kingdom Department of Mathematics and Computer Science Eindhoven University of Technology Netherlands Department of Computer Science University of Luxembourg Luxembourg

Sparse Neural Networks (SNNs) have emerged as powerful tools for efficient feature selection. Leveraging the dynamic sparse training (DST) algorithms within SNNs has demonstrated promising feature selection capabilities while drastically reducing computational overheads. Despite these advancements, several critical aspects remain insufficiently explored for feature selection. Questions persist regarding the choice of the DST algorithm for network training, the choice of metric for ranking features/neurons, and the comparative performance of these methods across diverse datasets when compared to dense networks. This paper addresses these gaps by presenting a comprehensive systematic analysis of feature selection with sparse neural networks. Moreover, we introduce a novel metric considering sparse neural network characteristics, which is designed to quantify feature importance within the context of SNNs. Our findings show that feature selection with SNNs trained with DST algorithms can achieve, on average, more than 50% memory and 55% FLOPs reduction compared to the dense networks, while outperforming them in terms of the quality of the selected features. Our code and the supplementary material are available on GitHub (https://***/zahraatashgahi/Neuron-Attribution). © 2024, CC BY-SA.

关键词： Feature Selection

来源：评论

学校读者我要写书评

暂无评论

HyperLeaf2024 – A Hyperspectral Imaging Dataset for Classification and Regression of Wheat Leaves

HyperLeaf2024 – A Hyperspectral Imaging Dataset for Classif...

引用

IEEE computer Society Conference on computer Vision and Pattern Recognition Workshops (CVPRW)

作者： William Michael Laprade Pawel Pieta Svetlana Kutuzova Jesper Cairo Westergaard Mads Nielsen Svend Christensen Anders Bjorholm Dahl Department of Applied Mathematics and Computer Science Technical University of Denmark Department of Computer Science University of Copenhagen Department of Plant and Environmental Sciences University of Copenhagen

ISBN: (数字)9798350365474

ISBN: (纸本)9798350365481

Hyperspectral imaging is a widely used method in remote sensing, particularly for use in airborne and satellite-based land surveillance. Its versatility is, however, much larger and has also seen usage in everything ranging from food processing and surveillance to astronomy and waste sorting. It is also gaining inroads with agricultural research. With most available datasets focusing on per-pixel classification, there is, however, a potential for hyperspectral whole-image analysis, but there is a severe lack of datasets for whole-image analysis. To help fill this gap and facilitate methodological development in whole-image hyperspectral image analysis, we introduce the Hy-perLeaf2024 dataset. The dataset consists of 2410 hyper-spectral images of wheat leaves, along with associated classification and regression targets at both the leaf level and the plot level. In addition to the dataset, we also provide experiments showing the importance of pretraining and highlighting the future research direction in whole-image hyper-spectral image analysis.

关键词： Adaptation models Image analysis Image coding Surveillance Conferences Focusing Distance measurement

来源：评论

学校读者我要写书评

暂无评论

Convolutional Neural Network with CBAM Module for Fitness Activity Recognition Using Wearable IMU Sensors 9

Convolutional Neural Network with CBAM Module for Fitness Ac...

引用

ISBN: (纸本)9798350318241

Wearable motion sensors are gaining traction for fitness activity tracking due to their utility in assisted living, wellness tracking, and fitness training applications. Such sensors typically incorporate accelerometers and gyroscopes to monitor and log physical activities. This work proposes a deep-learning approach for identifying fitness activities using wearable inertial measurement units (IMUs) signals. Specifically, a convolutional neural network (CNN) integrated with a convolutional block attention module (CBAM) provides the core framework. The CBAM module applies channel and spatial attention mechanisms to enhance CNN's capacity to discriminate salient features. Multichannel IMU time-series data, comprising accelerometer and gyroscope readings as subjects perform various exercises, supply input to the model. By learning via an attention-enhanced architecture, the model can extract robust and informative representations to categorize activities accurately. Experiments on a public IMU dataset with five exercise types across over twenty participants evaluated the approach. Results showed over 99.58% subject-independent activity recognition accuracy when augmenting the CNN with CBAM, constituting significant performance gains. The research findings indicate that an efficient deep-learning solution with attention modeling can recognize fitness activities reliably from on-body sensors. The proposed methodology holds substantial promise for applications in sports analytics and personalized health tracking. © 2024 IEEE.

关键词： Gyroscopes

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：