Pub Date : 2024-10-01DOI: 10.1109/ACCESS.2024.3464249
Issac Neha Margret;K. Rajakumar;K. V. Arulalan;S. Manikandan;Valentina E. Balas
Presents corrections to the paper, Corrections to “Statistical Insights Into Machine Learning-Based Box Models for Pregnancy Care and Maternal Mortality Reduction: A Literature Survey”.
对论文 "基于机器学习的妊娠护理和降低孕产妇死亡率方框模型的统计洞察 "进行更正：文献调查"。
{"title":"Corrections to “Statistical Insights Into Machine Learning-Based Box Models for Pregnancy Care and Maternal Mortality Reduction: A Literature Survey”","authors":"Issac Neha Margret;K. Rajakumar;K. V. Arulalan;S. Manikandan;Valentina E. Balas","doi":"10.1109/ACCESS.2024.3464249","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3464249","url":null,"abstract":"Presents corrections to the paper, Corrections to “Statistical Insights Into Machine Learning-Based Box Models for Pregnancy Care and Maternal Mortality Reduction: A Literature Survey”.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702422","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1109/ACCESS.2024.3461377
Xiaoqiang Zheng;Xinyu Ran;Mingxin Cai
{"title":"Retraction Notice: Short-Term Load Forecasting of Power System Based on Neural Network Intelligent Algorithm","authors":"Xiaoqiang Zheng;Xinyu Ran;Mingxin Cai","doi":"10.1109/ACCESS.2024.3461377","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3461377","url":null,"abstract":"","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702358","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-30DOI: 10.1109/ACCESS.2024.3469956
Junfeng Xie;Qingmin Jia;Youxing Chen
Advanced artificial intelligence (AI) and multi-access edge computing (MEC) technologies facilitate the development of edge intelligence, enabling the intelligence learned from remote cloud to network edge. To achieve automatic decision-making, the training efficiency and accuracy of AI models are crucial for edge intelligence. However, the collected data volume of each network edge node is limited, which may cause the over-fitting of AI models. To improve the training efficiency and accuracy of AI models for edge intelligence, intelligence networking-empowered edge computing (INEEC) is a promising solution, which enables each network edge node to improve its AI models quickly and economically with the help of other network edge nodes’ sharing of their learned intelligence. Sharing intelligence among network edge nodes efficiently is essential for INEEC. Thus in this paper, we study the intelligence sharing scheme, which aims to maximize the system energy efficiency while ensuring the latency tolerance via jointly optimizing intelligence requesting strategy, transmission power control and computation resource allocation. The system energy efficiency is defined as the ratio of model performance to energy consumption. Taking into account the dynamic characteristics of edge network conditions, the intelligence sharing problem is modeled as a Markov decision process (MDP). Subsequently, a twin delayed deep deterministic policy gradient (TD3)-based algorithm is designed to automatically make the optimal decisions. Finally, by extensive simulation experiments, it is shown that: 1) compared with DDPG and DQN, the proposed algorithm has a better convergence performance; 2) jointly optimizing intelligence requesting strategy, transmission power control and computation resource allocation helps to improve intelligence sharing efficiency; 3) under different parameter settings, the proposed algorithm achieves better results than the benchmark algorithms.
{"title":"Energy-Efficient Intelligence Sharing in Intelligence Networking-Empowered Edge Computing: A Deep Reinforcement Learning Approach","authors":"Junfeng Xie;Qingmin Jia;Youxing Chen","doi":"10.1109/ACCESS.2024.3469956","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3469956","url":null,"abstract":"Advanced artificial intelligence (AI) and multi-access edge computing (MEC) technologies facilitate the development of edge intelligence, enabling the intelligence learned from remote cloud to network edge. To achieve automatic decision-making, the training efficiency and accuracy of AI models are crucial for edge intelligence. However, the collected data volume of each network edge node is limited, which may cause the over-fitting of AI models. To improve the training efficiency and accuracy of AI models for edge intelligence, intelligence networking-empowered edge computing (INEEC) is a promising solution, which enables each network edge node to improve its AI models quickly and economically with the help of other network edge nodes’ sharing of their learned intelligence. Sharing intelligence among network edge nodes efficiently is essential for INEEC. Thus in this paper, we study the intelligence sharing scheme, which aims to maximize the system energy efficiency while ensuring the latency tolerance via jointly optimizing intelligence requesting strategy, transmission power control and computation resource allocation. The system energy efficiency is defined as the ratio of model performance to energy consumption. Taking into account the dynamic characteristics of edge network conditions, the intelligence sharing problem is modeled as a Markov decision process (MDP). Subsequently, a twin delayed deep deterministic policy gradient (TD3)-based algorithm is designed to automatically make the optimal decisions. Finally, by extensive simulation experiments, it is shown that: 1) compared with DDPG and DQN, the proposed algorithm has a better convergence performance; 2) jointly optimizing intelligence requesting strategy, transmission power control and computation resource allocation helps to improve intelligence sharing efficiency; 3) under different parameter settings, the proposed algorithm achieves better results than the benchmark algorithms.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10699330","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-30DOI: 10.1109/ACCESS.2024.3470754
Sang-Min Park;Jong-Eun Ha
3D semantic scene completion (SSC) aims to get a dense semantic understanding of an environment in 3D. It requires a geometric and semantic knowledge of the surrounding environment and the filling of void areas. In this paper, we propose an improved algorithm by modifying VoxFormer. VoxFormer consists of two steps for 3D semantic scene completion. First, it predicts the occupancy of an environment. Then, it completes the semantic scene completion through a masked autoencoder. It requires separate training for two stages, which can cause a disconnect of information from input to output. We propose an improved VoxFormer algorithm that makes end-to-end training possible by integrating occupancy prediction and scene completion. We use pseudo-LiDAR computed by depth estimation as input of 3D CNN, which generates queries for cross attention with 2D features. This makes the process end-to-end by connecting occupancy prediction and semantic scene completion. Experimental results using SemanticKITTI show improvement in the proposed algorithm.
三维语义场景补全（SSC）旨在获得对三维环境的密集语义理解。它需要周围环境的几何和语义知识以及空白区域的填充。在本文中，我们通过修改 VoxFormer 提出了一种改进算法。VoxFormer 包含两个三维语义场景补全步骤。首先，它预测环境的占用率。然后，它通过掩码自动编码器完成语义场景补全。它需要对两个阶段进行单独训练，这可能会造成从输入到输出的信息脱节。我们提出了一种改进的 VoxFormer 算法，通过整合占用预测和场景补全，使端到端的训练成为可能。我们使用通过深度估计计算出的伪激光雷达作为 3D CNN 的输入，而 3D CNN 会生成查询，以便与 2D 特征进行交叉关注。这就通过连接占用预测和语义场景补全实现了端到端的过程。使用 SemanticKITTI 的实验结果表明，所提出的算法有了很大改进。
{"title":"Semantic Scene Completion With 2D and 3D Feature Fusion","authors":"Sang-Min Park;Jong-Eun Ha","doi":"10.1109/ACCESS.2024.3470754","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3470754","url":null,"abstract":"3D semantic scene completion (SSC) aims to get a dense semantic understanding of an environment in 3D. It requires a geometric and semantic knowledge of the surrounding environment and the filling of void areas. In this paper, we propose an improved algorithm by modifying VoxFormer. VoxFormer consists of two steps for 3D semantic scene completion. First, it predicts the occupancy of an environment. Then, it completes the semantic scene completion through a masked autoencoder. It requires separate training for two stages, which can cause a disconnect of information from input to output. We propose an improved VoxFormer algorithm that makes end-to-end training possible by integrating occupancy prediction and scene completion. We use pseudo-LiDAR computed by depth estimation as input of 3D CNN, which generates queries for cross attention with 2D features. This makes the process end-to-end by connecting occupancy prediction and semantic scene completion. Experimental results using SemanticKITTI show improvement in the proposed algorithm.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10699323","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In pathology, various tissue and cell components play diverse biological roles. The morphology of each component can vary markedly with differentiation status or pathological conditions, making it critical for understanding diseases. Traditional computational pathology methods typically employ patch-based feature extraction, which aggregates visual features across entire images. However, this approach does not differentiate between tissue types, limiting component analysis. To address this limitation, we introduce a novel concept in pathology image analysis, namely segment representation learning, and present an algorithm, SegRep, for this purpose. SegRep uses a unique dual-masking strategy that combines input masking and feature map masking. This approach effectively removes external influences for the targeted segment, identified via a segmentation model or manual annotation, allowing for the extraction of segment-specific feature representations. In addition, SegRep utilizes a self-supervised learning algorithm to achieve optimized segment representation. We evaluated SegRep’s efficacy in clustering and classification tasks using a dataset of human gastric cancer samples. The results demonstrate SegRep’s superior capability in extracting feature vectors that are highly specific to different pathology image segments. Compared with traditional methods, SegRep shows significant improvements in accuracy and specificity in both clustering and classification tasks. Segment representations obtained via SegRep can offer a more detailed and insightful perspective on computational pathology, paving the way for advanced applications in the field.
{"title":"SegRep: Mask-Supervised Learning for Segment Representation in Pathology Images","authors":"Chichun Yang;Daisuke Komura;Mieko Ochi;Miwako Kakiuchi;Hiroto Katoh;Tetsuo Ushiku;Shumpei Ishikawa","doi":"10.1109/ACCESS.2024.3470213","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3470213","url":null,"abstract":"In pathology, various tissue and cell components play diverse biological roles. The morphology of each component can vary markedly with differentiation status or pathological conditions, making it critical for understanding diseases. Traditional computational pathology methods typically employ patch-based feature extraction, which aggregates visual features across entire images. However, this approach does not differentiate between tissue types, limiting component analysis. To address this limitation, we introduce a novel concept in pathology image analysis, namely segment representation learning, and present an algorithm, SegRep, for this purpose. SegRep uses a unique dual-masking strategy that combines input masking and feature map masking. This approach effectively removes external influences for the targeted segment, identified via a segmentation model or manual annotation, allowing for the extraction of segment-specific feature representations. In addition, SegRep utilizes a self-supervised learning algorithm to achieve optimized segment representation. We evaluated SegRep’s efficacy in clustering and classification tasks using a dataset of human gastric cancer samples. The results demonstrate SegRep’s superior capability in extracting feature vectors that are highly specific to different pathology image segments. Compared with traditional methods, SegRep shows significant improvements in accuracy and specificity in both clustering and classification tasks. Segment representations obtained via SegRep can offer a more detailed and insightful perspective on computational pathology, paving the way for advanced applications in the field.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10699328","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142377032","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-30DOI: 10.1109/ACCESS.2024.3470523
Sen Wang;Haiyang Wang;Chong Tang;Jiaxin Li;Daili Liang;Yanhua Qu
The sliding mode observer (SMO) has the advantages of small influence by parameter changes and strong robustness, which is widely used in the sensorless control of permanent magnet synchronous motor (PMSM). However, when the sliding mode observer is used to observe the position and speed information of the PMSM, the problem of slow response and excessive chattering is always accompanied. In order to reduce the time of observation position and improve the anti-interference of the system, the super-twisting sliding mode observer(STSMO) is proposed, which can be used to reduce the chattering amplitude effectively, Additionally, the fast and terminal factors is added to the sliding mode surface so that it can converge quickly in a finite time. Then, an improved fast terminal super-twisting sliding mode observer (FTSTSMO) is proposed in the extended state. It can be proved that the observer can converge by Lyapunov stability, and the new observer can suppress chattering and improve the convergence speed. Finally, the experimental analysis is carried out on the 1kW permanent magnet synchronous motor experimental platform, and the SMO and STSMO are compared with FTSTSMO. The results show that FTSTSMO can effectively reduce the fluctuation of the system, improve the tracking effect of rotor speed and rotor position when the speed changes, and further make the whole control system of PMSM have stronger robustness and stability.
{"title":"Research on Control Strategy of Permanent Magnet Synchronous Motor Based on Fast Terminal Super-Twisting Sliding Mode Observer","authors":"Sen Wang;Haiyang Wang;Chong Tang;Jiaxin Li;Daili Liang;Yanhua Qu","doi":"10.1109/ACCESS.2024.3470523","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3470523","url":null,"abstract":"The sliding mode observer (SMO) has the advantages of small influence by parameter changes and strong robustness, which is widely used in the sensorless control of permanent magnet synchronous motor (PMSM). However, when the sliding mode observer is used to observe the position and speed information of the PMSM, the problem of slow response and excessive chattering is always accompanied. In order to reduce the time of observation position and improve the anti-interference of the system, the super-twisting sliding mode observer(STSMO) is proposed, which can be used to reduce the chattering amplitude effectively, Additionally, the fast and terminal factors is added to the sliding mode surface so that it can converge quickly in a finite time. Then, an improved fast terminal super-twisting sliding mode observer (FTSTSMO) is proposed in the extended state. It can be proved that the observer can converge by Lyapunov stability, and the new observer can suppress chattering and improve the convergence speed. Finally, the experimental analysis is carried out on the 1kW permanent magnet synchronous motor experimental platform, and the SMO and STSMO are compared with FTSTSMO. The results show that FTSTSMO can effectively reduce the fluctuation of the system, improve the tracking effect of rotor speed and rotor position when the speed changes, and further make the whole control system of PMSM have stronger robustness and stability.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10699315","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-27DOI: 10.1109/ACCESS.2024.3469552
Zheng-Shao Chen;R. Vaitheeshwari;Eric Hsiao-Kuang Wu;Ying-Dar Lin;Ren-Hung Hwang;Po-Ching Lin;Yuan-Cheng Lai;Asad Ali
Advanced Persistent Threat (APT) groups pose significant cybersecurity threats due to their sophisticated and persistent nature. This study introduces a novel methodology to understand their collaborative patterns and shared objectives, which is crucial for developing robust defense mechanisms. We utilize MITRE ATT&CK Techniques, software, target nations, and industries as our primary features to understand the characteristics of APT groups. Since essential information is often buried within the unstructured data of Cyber Threat Intelligence (CTI) reports, we employ Natural Language Processing (NLP) and Named Entity Recognition (NER) to extract relevant data. To analyze and interpret the complex relationships between APT groups, we compute similarity among the features using weighted cosine similarity metrics and Machine Learning (ML) models, enhanced by feature crosses and feature selection strategies. Subsequently, hierarchical clustering is used to group APTs based on their similarity scores, helping to identify common behaviors and uncover deeper relationships. Our methodology demonstrates notable clustering performance, with a silhouette coefficient of 0.76, indicating strong intra-cluster similarity. The Adjusted Rand Index (ARI) of 0.63, though moderate, effectively measures agreement between our clustering and the ground truth. These metrics provide robust validation, surpassing commonly recognized benchmarks for effective clustering in cybersecurity. Our methodology successfully classifies 23 distinct APT groups into six clusters, highlighting the importance of techniques and industry features in the clustering process. Notably, techniques such as T1059 (Command and Scripting Interpreter) and T1036 (Masquerading) are prevalently deployed, observed in 18 out of 23 APT groups across all six clusters.
{"title":"Clustering APT Groups Through Cyber Threat Intelligence by Weighted Similarity Measurement","authors":"Zheng-Shao Chen;R. Vaitheeshwari;Eric Hsiao-Kuang Wu;Ying-Dar Lin;Ren-Hung Hwang;Po-Ching Lin;Yuan-Cheng Lai;Asad Ali","doi":"10.1109/ACCESS.2024.3469552","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3469552","url":null,"abstract":"Advanced Persistent Threat (APT) groups pose significant cybersecurity threats due to their sophisticated and persistent nature. This study introduces a novel methodology to understand their collaborative patterns and shared objectives, which is crucial for developing robust defense mechanisms. We utilize MITRE ATT&CK Techniques, software, target nations, and industries as our primary features to understand the characteristics of APT groups. Since essential information is often buried within the unstructured data of Cyber Threat Intelligence (CTI) reports, we employ Natural Language Processing (NLP) and Named Entity Recognition (NER) to extract relevant data. To analyze and interpret the complex relationships between APT groups, we compute similarity among the features using weighted cosine similarity metrics and Machine Learning (ML) models, enhanced by feature crosses and feature selection strategies. Subsequently, hierarchical clustering is used to group APTs based on their similarity scores, helping to identify common behaviors and uncover deeper relationships. Our methodology demonstrates notable clustering performance, with a silhouette coefficient of 0.76, indicating strong intra-cluster similarity. The Adjusted Rand Index (ARI) of 0.63, though moderate, effectively measures agreement between our clustering and the ground truth. These metrics provide robust validation, surpassing commonly recognized benchmarks for effective clustering in cybersecurity. Our methodology successfully classifies 23 distinct APT groups into six clusters, highlighting the importance of techniques and industry features in the clustering process. Notably, techniques such as T1059 (Command and Scripting Interpreter) and T1036 (Masquerading) are prevalently deployed, observed in 18 out of 23 APT groups across all six clusters.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10697172","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376574","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-27DOI: 10.1109/ACCESS.2024.3469197
Onur N. Tepencelik;Wenchuan Wei;Pamela C. Cosman;Sujit Dey
We propose a system that estimates people’s body and head orientations using low-resolution point cloud data from two LiDAR sensors. Our models make accurate estimations in real-world conversation settings where subjects move naturally with varying head and body poses, while seated around a table. The body orientation estimation model uses ellipse fitting while the head orientation estimation model combines geometric feature extraction with an ensemble of neural network regressors. Our models achieve a mean absolute estimation error of 5.2 degrees for body orientation and 13.7 degrees for head orientation. Compared to other body/head orientation estimation systems that use RGB cameras, our proposed system uses LiDAR sensors to preserve user privacy, while achieving comparable accuracy. Unlike other body/head orientation estimation systems, our sensors do not require a specified close-range placement in front of the subject, enabling estimation from a surveillance viewpoint which produces low-resolution data. This work is the first to attempt head orientation estimation using point clouds in a low-resolution surveillance setting. We compare our model to two state-of-the-art head orientation estimation models that are designed for high-resolution point clouds, which yield higher estimation errors on our low-resolution dataset. We also present an application of head orientation estimation by quantifying behavioral differences between neurotypical and autistic individuals in triadic (three-way) conversations. Significance tests show that autistic individuals display significantly different behavior compared to neurotypical individuals in distributing attention between conversational parties, suggesting that the approach could be a component of a behavioral analysis or coaching system.
{"title":"Body and Head Orientation Estimation From Low-Resolution Point Clouds in Surveillance Settings","authors":"Onur N. Tepencelik;Wenchuan Wei;Pamela C. Cosman;Sujit Dey","doi":"10.1109/ACCESS.2024.3469197","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3469197","url":null,"abstract":"We propose a system that estimates people’s body and head orientations using low-resolution point cloud data from two LiDAR sensors. Our models make accurate estimations in real-world conversation settings where subjects move naturally with varying head and body poses, while seated around a table. The body orientation estimation model uses ellipse fitting while the head orientation estimation model combines geometric feature extraction with an ensemble of neural network regressors. Our models achieve a mean absolute estimation error of 5.2 degrees for body orientation and 13.7 degrees for head orientation. Compared to other body/head orientation estimation systems that use RGB cameras, our proposed system uses LiDAR sensors to preserve user privacy, while achieving comparable accuracy. Unlike other body/head orientation estimation systems, our sensors do not require a specified close-range placement in front of the subject, enabling estimation from a surveillance viewpoint which produces low-resolution data. This work is the first to attempt head orientation estimation using point clouds in a low-resolution surveillance setting. We compare our model to two state-of-the-art head orientation estimation models that are designed for high-resolution point clouds, which yield higher estimation errors on our low-resolution dataset. We also present an application of head orientation estimation by quantifying behavioral differences between neurotypical and autistic individuals in triadic (three-way) conversations. Significance tests show that autistic individuals display significantly different behavior compared to neurotypical individuals in distributing attention between conversational parties, suggesting that the approach could be a component of a behavioral analysis or coaching system.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10697132","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142377090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-27DOI: 10.1109/ACCESS.2024.3469912
Qiong Wu;Hua Chen;Baolong Liu
Fuel cell vehicles have rapidly occupied the market with advantages such as environmental protection and energy conservation. However, their battery technology is insufficient and their endurance is poor, making them unsuitable for use over long distances. To address the aforementioned issues, a fuel cell vehicle energy storage system based on super-capacitors was constructed. Meanwhile, a proportional integral derivative controller based on fuzzy algorithms was established. Finally, the particle swarm optimization algorithm was used to optimize the fuzzy control strategy that integrated the fuzzy algorithm. When using the optimized fuzzy control strategy for simulation, the peak power of the fuel cell output power was reduced from 3.8kW to 2.0kW. The remaining power of the super-capacitor remained stable within a reasonable range throughout the entire operating condition. Under the new European urban road cycle, the optimized control strategy improved energy recovery performance by 4.3% and reduced hydrogen consumption by 0.9964%. Under the United States federal environmental protection agency standardized urban cycle conditions, the optimized control strategy improved the braking energy recovery efficiency index and effective braking energy recovery efficiency by 8.9% and 6.3%, respectively. The percentage reduction in hydrogen consumption was 0.9433%. Therefore, this research method can effectively reduce hydrogen consumption and improve the product economy and market competitiveness of enterprises.
{"title":"Optimization of Control Strategy for Fuel Cell Vehicles by Integrating Fuzzy Algorithm","authors":"Qiong Wu;Hua Chen;Baolong Liu","doi":"10.1109/ACCESS.2024.3469912","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3469912","url":null,"abstract":"Fuel cell vehicles have rapidly occupied the market with advantages such as environmental protection and energy conservation. However, their battery technology is insufficient and their endurance is poor, making them unsuitable for use over long distances. To address the aforementioned issues, a fuel cell vehicle energy storage system based on super-capacitors was constructed. Meanwhile, a proportional integral derivative controller based on fuzzy algorithms was established. Finally, the particle swarm optimization algorithm was used to optimize the fuzzy control strategy that integrated the fuzzy algorithm. When using the optimized fuzzy control strategy for simulation, the peak power of the fuel cell output power was reduced from 3.8kW to 2.0kW. The remaining power of the super-capacitor remained stable within a reasonable range throughout the entire operating condition. Under the new European urban road cycle, the optimized control strategy improved energy recovery performance by 4.3% and reduced hydrogen consumption by 0.9964%. Under the United States federal environmental protection agency standardized urban cycle conditions, the optimized control strategy improved the braking energy recovery efficiency index and effective braking energy recovery efficiency by 8.9% and 6.3%, respectively. The percentage reduction in hydrogen consumption was 0.9433%. Therefore, this research method can effectively reduce hydrogen consumption and improve the product economy and market competitiveness of enterprises.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10697117","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376708","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-27DOI: 10.1109/ACCESS.2024.3469155
Hasindu Dewasurendra;Taejoon Kim
Despite extensive applications in surveillance and remote sensing, research on very low-resolution (VLR) image classification remains relatively unexplored in comparison to high-resolution (HR) image classification. We introduce a deep hybrid network that integrates capsule routing networks with a two-layer attention module. In the proposed architecture, the attention mechanism captures the more salient features, and the capsule network encodes these features to be robust to resolution changes. To enhance the network’s performance, a transfer learning on a custom image dataset, which is well-aligned to CIFAR-10, is utilized. The proposed model (Codes for the models are available at: https://github.com/kdhasi/Deep-CapsuleAttention.git ) is evaluated on two VLR classification tasks of ‘VLR complex image’ and ‘VLR real-world digit’. Experimental results demonstrate the superiority of the proposed model, achieving state-of-the-art (SOTA) results in both VLR complex image and VLR real-world digit domains while using fewer parameters compared to previous SOTA networks. Specifically, on the VLR CIFAR-10 dataset, the proposed model attains a 3.17% improvement in detection accuracy over the current benchmarks, and, on the VLR SVHN dataset, it achieves a 3.85% improvement by using 80% fewer parameters.
{"title":"Deep Hybrid Architecture for Very Low-Resolution Image Classification Using Capsule Attention","authors":"Hasindu Dewasurendra;Taejoon Kim","doi":"10.1109/ACCESS.2024.3469155","DOIUrl":"https://doi.org/10.1109/ACCESS.2024.3469155","url":null,"abstract":"Despite extensive applications in surveillance and remote sensing, research on very low-resolution (VLR) image classification remains relatively unexplored in comparison to high-resolution (HR) image classification. We introduce a deep hybrid network that integrates capsule routing networks with a two-layer attention module. In the proposed architecture, the attention mechanism captures the more salient features, and the capsule network encodes these features to be robust to resolution changes. To enhance the network’s performance, a transfer learning on a custom image dataset, which is well-aligned to CIFAR-10, is utilized. The proposed model (Codes for the models are available at: \u0000<uri>https://github.com/kdhasi/Deep-CapsuleAttention.git</uri>\u0000) is evaluated on two VLR classification tasks of ‘VLR complex image’ and ‘VLR real-world digit’. Experimental results demonstrate the superiority of the proposed model, achieving state-of-the-art (SOTA) results in both VLR complex image and VLR real-world digit domains while using fewer parameters compared to previous SOTA networks. Specifically, on the VLR CIFAR-10 dataset, the proposed model attains a 3.17% improvement in detection accuracy over the current benchmarks, and, on the VLR SVHN dataset, it achieves a 3.85% improvement by using 80% fewer parameters.","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":null,"pages":null},"PeriodicalIF":3.4,"publicationDate":"2024-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10697178","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142376973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：