检索结果-内蒙古大学图书馆

World Congress on Computing and Communication Technologies (WCCCT)

作者： Zijing Rong Yaowen Liu Mingjun Dai Stefano Savazzi Sanaz Kianoush College of Electronic and Information Engineering Shenzhen University Shenzhen China School of Computer Science and Technology Kashi University China Institute of Electronics Information Engineering and Telecommunications (IEIIT) National Research Council Italy IEIIT National Research Council Italy

ISBN: (数字)9798331512620

ISBN: (纸本)9798331512637

The FedAdamPID algorithm resolves the overshoot problem suffered by the FedAdam algorithm under the framework of PID control. The research on FedAdamPID explores the connection between PID controllers and the process of federated learning algorithms, particularly applying it to the update rules on the server side. This study draws on the previous research of the research group, first establishing a connection between the local update process of the client and the PID controller, and then viewing the entire FL process involving both the client and the server as a multi-PID coupled system. On this basis, the minimum update difference and the dynamic adjustment strategy of the local learning rate are introduced, and an optimized algorithm, FedAdamPID-op, is proposed. Experiments on multiple standard datasets have verified that this optimized algorithm slightly outperforms FedAdamPID in terms of accuracy and the number of convergence rounds.

关键词： Adaptive learning Adaptation models Accuracy Federated learning Heuristic algorithms Process control Minimization Servers Standards Convergence

来源：评论

学校读者我要写书评

暂无评论

pFedLVM: A Large Vision Model (LVM)-Driven and Latent Feature-Based Personalized Federated Learning Framework in Autonomous Driving

arXiv

引用

arXiv 2024年

作者： Kou, Wei-Bin Lin, Qingfeng Tang, Ming Xu, Sheng Ye, Rongguang Leng, Yang Wang, Shuai Li, Guofa Chen, Zhenyu Zhu, Guangxu Wu, Yik-Chung Department of Electrical and Electronic Engineering The University of Hong Kong Hong Kong Shenzhen Research Institute of Big Data Shenzhen China Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen China Research Institute of Electronic Science and Technology University of Electronic Science and technology Chengdu China Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China State Key Laboratory for Novel Software Technology Nanjing University Nanjing China College of Mechanical and Vehicle Engineering Chongqing University Chongqing China

Deep learning-based Autonomous Driving (AD) semantic segmentation (SSeg) models often exhibit poor generalization due to data heterogeneity in an ever domain-shifting environment. While Federated Learning (FL) could improve the generalization of an AD SSeg model (known as FedAD system), conventional models often struggle with under-fitting as the amount of accumulated training data progressively increases. To address this issue, instead of conventional small models, employing Large Vision Models (LVMs) in FedAD is a viable option for better learning of representations from a vast volume of data. However, implementing LVMs in FedAD introduces three challenges: (I) the extremely high communication overheads associated with transmitting LVMs between participating vehicles and a central server;(II) lack of computing resource to deploy LVMs on each vehicle;(III) the performance drop due to LVM focusing on shared features but overlooking local vehicle characteristics. To overcome these challenges, we propose pFedLVM, a LVM-Driven, Latent Feature-Based Personalized Federated Learning framework. In this approach, the LVM is deployed only on central server, which effectively alleviates the computational burden on individual vehicles. Furthermore, the exchange between central server and vehicles are the learned features rather than the LVM parameters, which significantly reduces communication overhead. In addition, we utilize both shared features from all participating vehicles and individual characteristics from each vehicle to establish a personalized learning mechanism. This enables each vehicle’s model to learn features from others while preserving its personalized characteristics, thereby outperforming globally shared models trained in general FL. Extensive experiments demonstrate that pFedLVM outperforms the existing state-of-the-art approach by 18.47%, 25.60%, 51.03% and 14.19% in terms of mIoU, mF1, mPrecision and mRecall, respectively. © 2024, CC BY-NC-ND.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

ToC-Net: Combining Channel Attention for Targeted Optimization of Stroke Segmentation Networks

ToC-Net: Combining Channel Attention for Targeted Optimizati...

引用

Medical Artificial Intelligence (MedAI), IEEE International Conference on

作者： Yazhou Li Yang Li Yan Zhou Computer Science and Engineering Changchun University of Technology Changchun China Frontier cross research institute Northeast Normal University Changchun China

Ischemic stroke is a common cerebrovascular disease, and rapid and accurate identification of stroke lesions in patients to assess risk and obtain the most timely and effective treatment can reduce long-term disability and mortality. Therefore, it is clinically important to study stroke segmentation methods. This paper propose a combining channel attention for targeted optimization of stroke segmentation networks (ToC-Net) to achieve accurate segmentation of 2D stroke lesions by addressing the problems of vague lesion contours and insignificant intra-class differences. U-Net++ is used as the baseline network, and the channel attention module with weights (WCA) and atrous spatial pyramid pooling module are added in the feature extraction part to unify the feature map size of each channel, highlight the lesion information in different scales, and then extract the global features. Finally, the network output is corrected by the target calibration (TC) module to obtain the final segmentation results. This paper compares with five state-of-the-art methods on the public dataset ISLES2022 and private dataset (AIS), and the results validate that ToC-Net can segment stroke lesion regions effectively and the segmentation results are better than the comparison network. The project website can be found at https://***/lucky-asia/ToC-Net.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Topological Characteristics of 5d Spatially Dynamic Brain Networks in Schizophrenia

Topological Characteristics of 5d Spatially Dynamic Brain Ne...

引用

IEEE International Symposium on Biomedical Imaging

作者： Mustafa S. Salman Armin Iraji Noah Lewis Vince D. Calhoun School of Electrical and Computer Engineering Georgia Institute of Technology Tri-Institutional Center for Translational Research in Neuroimaging & Data Science (TReNDS) Georgia Institute of Technology Georgia State University & Emory University School of Computational Science and Engineering Georgia Institute of Technology Atlanta GA USA

The last decade of rich data-driven research on functional magnetic resonance imaging (fMRI) has provided novel insights into human brain function and aberrant behavior in brain disorders. Independent component analysis (ICA) is a widely-used technique for data-driven analysis of fMRI data. Spatial ICA is the most prominent variation of ICA and provides replicable and interpretable intrinsic connectivity network (ICN). It assumes common spatial activation across time. However, very recent studies indicate that there is utility in adopting a dynamic spatial activation modeling approach. Characterizing dynamics for both temporal and spatial domains means we have a multitude of decompositions of the already high-dimensional, multi-dataset, and multi-subject fMRI data. Hence making sense of the derived data becomes a significant issue. Here we use topological data analysis (TDA) to identify topological descriptors of the windowed spatially dynamic components of fMRI data. We discover and summarize differences in the spatial dynamics of controls and schizophrenia patients (SZs). We discover that SZs generally have lower Betti numbers and higher Wasserstein distance between spatiotemporal brain states, which provide intuitive summaries of the reduced dynamism SZs exhibit in resting-state fMRI studies.

关键词：

来源：评论

学校读者我要写书评

暂无评论

DuDoUniNeXt: Dual-domain unified hybrid model for single and multi-contrast undersampled MRI reconstruction

arXiv

引用

arXiv 2024年

作者： Gao, Ziqi Zhang, Yue Liu, Xinwen Li, Kaiyan Zhou, S. Kevin School of Biomedical Engineering Division of Life Sciences and Medicine University of Science and Technology of China Anhui Hefei230026 China Suzhou Institute for Advance Research University of Science and Technology of China Jiangsu Suzhou215123 China School of Electrical Engineering and Computer Science The University of Queensland Australia

Multi-contrast (MC) Magnetic Resonance Imaging (MRI) reconstruction aims to incorporate a reference image of auxiliary modality to guide the reconstruction process of the target modality. Known MC reconstruction methods perform well with a fully sampled reference image, but usually exhibit inferior performance, compared to single-contrast (SC) methods, when the reference image is missing or of low quality. To address this issue, we propose DuDoUniNeXt, a unified dual-domain MRI reconstruction network that can accommodate to scenarios involving absent, low-quality, and high-quality reference images. DuDoUniNeXt adopts a hybrid backbone that combines CNN and ViT, enabling specific adjustment of image domain and k-space reconstruction. Specifically, an adaptive coarse-to-fine feature fusion module (AdaC2F) is devised to dynamically process the information from reference images of varying qualities. Besides, a partially shared shallow feature extractor (PaSS) is proposed, which uses shared and distinct parameters to handle consistent and discrepancy information among contrasts. Experimental results demonstrate that the proposed model surpasses state-of-the-art SC and MC models significantly. Ablation studies show the effectiveness of the proposed hybrid backbone, AdaC2F, PaSS, and the dual-domain unified learning scheme. © 2024, CC BY.

关键词： Magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

A Classification Model for Pesticide Residues in Food Based On Machine Learning Methods

A Classification Model for Pesticide Residues in Food Based ...

引用

Innovative Trends in Information technology (ICITIIT), International Conference on

作者： Nayani Sateesh Navneet Rawat Suraj Shankarrao Damre Warish Patel Harshal Patil Ramya Maranan Department of Computer Science and Engineering CVR College of Engineering Hyderabad Telangana India Department of Management Graphic Era Deemed to be University Dehradun Uttarakhand India Department of Information Technology D Y Patil College of Engineering Akurdi Pune Maharashtra India Department of Computer Science and Engineering Parul Institute of Engineering and Technology Parul University Post Limda Waghodia Gujarat India Department of Computer Science and Engineering Symbiosis Institute of Technology Symbiosis International (Deemed University) Pune Maharashtra India Department of Research and Innovation Saveetha School of Engineering SIMATS Chennai Tamil Nadu India

ISBN: (数字)9798350386813

ISBN: (纸本)9798350386820

The increasing concern over pesticide residues in food necessitates the development of accurate and efficient tools for residue detection and classification. This research proposes a novel machine learning-based classification model for predicting the presence and types of pesticide residues in food items. Leveraging advanced data analysis techniques, the model aims to enhance food safety measures and facilitate regulatory compliance. The classification model exhibits promising results in accurately identifying the presence of pesticide residues in food items. The chosen machine learning algorithms showcase distinct strengths in handling the complexity of pesticide residue patterns across different food categories. Based on machine learning methods, the classification model performs with an error of 1.03%. A model has been designed that makes use of three machine learning methods, choosing the most appropriate one, which is capable of analyzing 130,431 food data and classifying them, using it for new food data with possible high amounts of pesticide residues.

关键词： Analytical models Machine learning algorithms Pesticides Machine learning Data models Sensor systems Numerical models

来源：评论

学校读者我要写书评

暂无评论

Shielding NLP Systems: An In-depth Survey on Advanced AI Techniques for Adversarial Attack Detection in Cyber Security

Shielding NLP Systems: An In-depth Survey on Advanced AI Tec...

引用

Automation, Computing and Renewable Systems (ICACRS), International Conference on

作者： T. Sai Lalith Prasad K. B. Manikandan J. Vinoj Research Scholar Department of Computer Science and Engineering Vignan's Foundation for Science and Technology & Research (Deemed to be University) Guntur Andhra Pradesh India Assistant Professor Department of Artificial Intelligence and Data Science Vignan Institute of Technology and Science Telangana India Assistant Professor Department of CSE Vignan's Foundation for Science Technology and Research (Deemed to be University) Guntur Andhra Pradesh India

ISBN: (数字)9798331532420

ISBN: (纸本)9798331532437

This research investigates the challenges posed by adversarial perturbations to the robustness and reliability of Natural Language Processing (NLP) models. The study focuses on three key paradigms for enhancing the resilience of NLP systems: meta-learning, reinforcement learning, and self-supervised learning. A comprehensive evaluation of these strategies is conducted, analyzing their strengths and weaknesses in mitigating adversarial attacks. The study provides insights into the most promising approaches for enhancing the robustness of NLP models and offers directions for future research in this critical area. By advancing the resilience of NLP systems, this research contributes to the development of more robust and trustworthy AI applications across various domains, including healthcare, finance, and security.

关键词： Training Surveys Renewable energy sources Scalability Self-supervised learning Natural language processing Robustness Security Artificial intelligence Resilience

来源：评论

学校读者我要写书评

暂无评论

Smart Controller for Air Conditioning in Car using IoT

Smart Controller for Air Conditioning in Car using IoT

引用

Electronics and Renewable Systems (ICEARS), International Conference on

作者： Pradeepkumar G Manikandan P Rathika P Senthil T Thurai Raaj V B Ponmurugan P Department of Electronics and Communication Engineering KPR Institute of Engineering and Technology Coimbatore India Department of Electrical and Electronics Engineering K S R Institute for Engineering and Technology Tiruchengode Department of Computer Science and Engineering K S R Institute for Engineering and Technology Tiruchengode Department of Electronics and Communication Engineering Bannari Amman Institute of Technology Sathyamangalam Tamilnadu Department of Electrical and Electronics Engineering Madanapalle Institute of Technology & Science Madanapalle Andhra Pradesh Department of Research Meenakshi Academy of Higher Education and Research Chennai

The vast majority of automobiles now have intelligent controllers. During the hot summer months, the interior temperature of the car rapidly rises, necessitating the use of the air conditioner to maintain comfort. This necessitates the need to turn ON the AC before getting into the car. The proposed mobile application is mainly developed to monitor the temperature range inside the car by using the Internet of Things (IoT). With the proposed mobile application, the air conditioning can be switched ON before getting into the vehicle as the AC controller is linked to the mobile app. On the other hand, it is highly required to ensure that the air conditioner has sufficient power. This necessitates the use of a lithium-ion battery. Electrical energy is created by converting kinetic energy. The wingtips will generate kinetic energy as the vehicle moves. The intelligent air conditioner's energy consumption will decrease as the battery is charged. The results show that the proposed system can contribute significantly to the proposed temperature reduction. The proposed strategy will monitor the temperature inside the vehicle as simple as possible.

关键词： Temperature sensors Temperature measurement Mood Control systems Mobile applications Temperature control Automobiles

来源：评论

学校读者我要写书评

暂无评论

Deep Omni-supervised Learning for Rib Fracture Detection from Chest Radiology Images

arXiv

引用

arXiv 2023年

作者： Chai, Zhizhong Luo, Luyang Lin, Huangjing Heng, Pheng-Ann Chen, Hao Imsight AI Research Lab Shenzhen China The Department of Computer Science and Engineering The Hong Kong University of Science and Technology Hong Kong The Department of Computer Science and Engineering Department of Chemical and Biological Engineering Hong Kong University of Science and Technology Hong Kong HKUST Shenzhen-Hong Kong Collaborative Innovation Research Institute Futian Shenzhen China The Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong

Deep learning (DL)-based rib fracture detection has shown promise of playing an important role in preventing mortality and improving patient outcome. Normally, developing DL-based object detection models requires a huge amount of bounding box annotation. However, annotating medical data is time-consuming and expertise-demanding, making obtaining a large amount of fine-grained annotations extremely infeasible. This poses a pressing need for developing label-efficient detection models to alleviate radiologists’ labeling burden. To tackle this challenge, the literature on object detection has witnessed an increase of weakly-supervised and semi-supervised approaches, yet still lacks a unified framework that leverages various forms of fully-labeled, weakly-labeled, and unlabeled data. In this paper, we present a novel omni-supervised object detection network, ORF-Netv2, to leverage as much available supervision as possible. Specifically, a multi-branch omni-supervised detection head is introduced with each branch trained with a specific type of supervision. A co-training-based dynamic label assignment strategy is then proposed to enable flexible and robust learning from the weakly-labeled and unlabeled data. Extensive evaluation was conducted for the proposed framework with three rib fracture datasets on both chest CT and X-ray. By leveraging all forms of supervision, ORF-Netv2 achieves mAPs of 34.7, 44.7, and 19.4 on the three datasets, respectively, surpassing the baseline detector which uses only box annotations by mAP gains of 3.8, 4.8, and 5.0, respectively. Furthermore, ORF-Netv2 consistently outperforms other competitive label-efficient methods over various scenarios, showing a promising framework for label- © 2023, CC BY.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

AI-Enhanced Living: The Future of Smart Homes

AI-Enhanced Living: The Future of Smart Homes

引用

Smart Devices (ICSD), International Conference on

作者： Aashna Sinha Nishesh Sharma Shivam Kumar Ayapal Lande Mohammed Ismail Iqbal School of Applied and Life Science Division of Research & Innovation Uttaranchal University Dehradun India Aerospace Engineering Uttaranchal Institute of Technology Uttaranchal University Dehradun India School of Computer Science & Artificial Intelligence SR University Warangal College of Engineering and Technology University of Technology and Sciences Nizwa

ISBN: (数字)9798350394412

ISBN: (纸本)9798350394429

An important area of attention is energy efficiency since artificial intelligence (AI) algorithms may control power use by identifying occupant habits and modifying systems accordingly. This not only lessens the impact on the environment but also generates significant financial savings over time. A new age of ease, effectiveness, and security for homeowners has arrived with the integration of AI in smart homes. Machine learning algorithms are used in AI-powered smart homes to automate and improve a variety of daily tasks. These solutions, which range from voice-activated virtual assistants to intelligent lighting and climate management, adapt and react to the particular tastes and demands of residents. AI-driven surveillance substantially improves security by utilizing tools like face recognition, predictive analytics, and remote monitoring to provide users piece of mind. Smart homes powered by AI provide advantages in terms of healthcare and wellbeing in addition to convenience and efficiency. Modern home-based health monitoring devices may measure vital signs and give early warnings of potential health problems. Reminders for taking medications and assistance for seniors who want to age in place can improve quality of life. A unified environment that adjusts and anticipates inhabitant demands is created via interconnected gadgets and central control centers. Strong data protection and secure network protocols are necessary since cybersecurity and privacy are urgent issues. The key developments and future ramifications of AI-driven smart homes are summarized in this study

关键词： Machine learning algorithms Protocols Virtual assistants Surveillance Smart homes Aerospace electronics Artificial intelligence Sustainable development Smart devices Remote monitoring

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：