检索结果-内蒙古大学图书馆

International Conference on Computer Vision (ICCV)

作者： Haozhe Liu Haoqian Wu Weicheng Xie Feng Liu Linlin Shen Computer Vision Institute College of Computer Science and Software Engineering SZU Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society National Engineering Laboratory for Big Data System Computing Technology Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen China

The convolutional neural network (CNN) is vulnerable to degraded images with even very small variations (e.g. corrupted and adversarial samples). One of the possible reasons is that CNN pays more attention to the most discriminative regions, but ignores the auxiliary features when learning, leading to the lack of feature diversity for final judgment. In our method, we propose to dynamically suppress significant activation values of CNN by group-wise inhibition, but not fixedly or randomly handle them when training. The feature maps with different activation distribution are then processed separately to take the feature independence into account. CNN is finally guided to learn richer discriminative features hierarchically for robust classification according to the proposed regularization. Our method is comprehensively evaluated under multiple settings, including classification against corruptions, adversarial attacks and low data regime. Extensive experimental results show that the proposed method can achieve significant improvements in terms of both robustness and generalization performances, when compared with the state-of-the-art methods. Code is available at https://***/LinusWu/TENET_Training.

关键词： Training Computer vision Codes Heuristic algorithms Robustness Classification algorithms Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

PhysLLM: Harnessing Large Language Models for Cross-Modal Remote Physiological Sensing

arXiv

引用

arXiv 2025年

作者： Xie, Yiping Zhao, Bo Dai, Mingtong Zhou, Jian-Ping Sun, Yue Tan, Tao Xie, Weicheng Shen, Linlin Yu, Zitong Shenzhen University China Great Bay University China National Engineering Laboratory for Big Data System Computing Technology China Dongguan Key Laboratory for Intelligence and Information Technology China Guangdong Medical University China Southern Medical University Dongguan People’s Hospital China Macao Polytechnic University China

Remote photoplethysmography (rPPG) enables noncontact physiological measurement but remains highly susceptible to illumination changes, motion artifacts, and limited temporal modeling. Large Language Models (LLMs) excel at capturing long-range dependencies, offering a potential solution but struggle with the continuous, noise-sensitive nature of rPPG signals due to their text-centric design. To bridge this gap, we introduce PhysLLM, a collaborative optimization framework that synergizes LLMs with domain-specific rPPG components. Specifically, the Text Prototype Guidance (TPG) strategy is proposed to establish cross-modal alignment by projecting hemodynamic features into LLM-interpretable semantic space, effectively bridging the representational gap between physiological signals and linguistic tokens. Besides, a novel Dual-Domain Stationary (DDS) Algorithm is proposed for resolving signal instability through adaptive time-frequency domain feature re-weighting. Finally, rPPG task-specific cues systematically inject physiological priors through physiological statistics, environmental contextual answering, and task description, leveraging cross-modal learning to integrate both visual and textual information, enabling dynamic adaptation to challenging scenarios like variable illumination and subject movements. Evaluation on four benchmark datasets, PhysLLM achieves state-of-the-art accuracy and robustness, demonstrating superior generalization across lighting variations and motion scenarios. Copyright © 2025, The Authors. All rights reserved.

关键词： Physiological models

来源：评论

学校读者我要写书评

暂无评论

Deep Learning for Unsupervised Anomaly Localization in Industrial Images: A Survey

arXiv

引用

arXiv 2022年

作者： Tao, Xian Gong, Xinyi Zhang, Xin Yan, Shaohua Adak, Chandranath The Research Center of Precision Sensing and Control Institute of Automation Chinese Academy of Sciences Beijing100190 China The School of Artificial Intelligence University of Chinese Academy of Sciences Beijing100049 China Key Laboratory of Industrial Internet and Big Data China National Light Industry Beijing Technology and Business University Beijing100048 China The Dept. of CSE Indian Institute of Technology Bihar Patna801106 India

Currently, deep learning-based visual inspection has been highly successful with the help of supervised learning methods. However, in real industrial scenarios, the scarcity of defect samples, the cost of annotation, and the lack of a priori knowledge of defects may render supervised-based methods ineffective. In recent years, unsupervised anomaly localization algorithms have become more widely used in industrial inspection tasks. This paper aims to help researchers in this field by comprehensively surveying recent achievements in unsupervised anomaly localization in industrial images using deep learning. The survey reviews more than 120 significant publications covering different aspects of anomaly localization, mainly covering various concepts, challenges, taxonomies, benchmark datasets, and quantitative performance comparisons of the methods reviewed. In reviewing the achievements to date, this paper provides detailed predictions and analysis of several future research directions. This review provides detailed technical information for researchers interested in industrial anomaly localization and who wish to apply it to the localization of anomalies in other fields. Copyright © 2022, The Authors. All rights reserved.

关键词： Defects

来源：评论

学校读者我要写书评

暂无评论

Roof greening in major Chinese cities possibly afford a large potential carbon sink

引用

Science Bulletin 2024年第20期69卷 3216-3220页

作者： Chao Yang Yinghui Zhang Min Chen Song Zhu Yuzhi Tang Zhixin Zhang Wei Ma Huizeng Liu Junyi Chen Bohui Tang Dejin Zhang Zhengdong Huang Xuqing Wang Wei Tu Cuiling Liu Tiezhu Shi Haiying Xu Aihong Cui Fanyi Meng Tianhong Zhao Kai Guo Wenhao Guo Bi Fan Quande Qin Wuyang Hong Haishan Wu Bing Wang Jinrong Zeng Ka Wai Wu Qingquan Li Key Laboratory for Geo-Environmental Monitoring of Great Bay Area Ministry of Natural Resources&Guangdong Key Laboratory of Urban InformaticsShenzhen UniversityShenzhen 518060China School of Architecture and Urban Planning Shenzhen UniversityShenzhen 518060China Key Laboratory of Virtual Geographic Environment Ministry of EducationNanjing Normal UniversityNanjing 210023China Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application Nanjing 210023China International Research Center of Big Data for Sustainable Development Goals Beijing 100094China Shenzhen Zhiyuan Space Innovation Technology Co. LtdShenzhen 518060China Guangdong Laboratory of Artificial Intelligence and Digital Economy(SZ) Shenzhen 518107China School of Geography and Ocean Science Nanjing UniversityNanjing 210008China School of Geography Nanjing Normal UniversityNanjing 210023China School of Civil Engineering Chongqing Jiaotong UniversityChongqing 400074China Institute for Advanced Study&Shenzhen University Shenzhen 518060China Tiandu-Shenzhen University Deep Space Joint Laboratory Shenzhen UniversityShenzhen 518060China Faculty of Land Resource Engineering Kunming University of Science and TechnologyKunming 650093China Center for Hydrogeology and Environmental Geology China Geological SurveyBaoding 071051China Department of Landscape Architecture and Urban Planning Texas A&M UniversityTX 77840USA Department of Geography Hong Kong Baptist UniversityHong Kong 999077China College of Civil and Transportation Engineering Shenzhen UniversityShenzhen 518060China College of Big Data and Internet Shenzhen Technology UniversityShenzhen 518118China Faculty of Geosciences and Environmental Engineering Southwest Jiaotong UniversityChengdu 611756China College of Management Shenzhen UniversityShenzhen 518060China WeBank(Shenzhen)Co. LtdShenzhen 518060China Department of Aeronautical and Aviation Engineering The Hong Kong Polytechnic UniversityHong Kong 100872China Esri C

China is experiencing accelerated urbanisation,with a large number of people moving from rural to urban areas[1].It has resulted in large losses in the net primary production(NPP),biodiversity and carbon stocks and an increase in environmental pollution and CO_(2)emissions[2–4].In 2015,196 countries signed the Paris Agreement and committed to setting long-term goals to jointly manage climate change and reduce their individual emissions,aiming to control the increase in global average temperature from the pre-industrial level to below 2℃and to curtail the temperature rise within 1.5℃till the end of the 21st century[5].China is bolstering its efforts to achieve the climate change mitigation goals and has announced a plan for achieving carbon neutrality by 2060[6].The carbon neutrality goal poses a challenge to the current policies promoting rapid urbanisation across China.

关键词： jointly carbon goals

来源：评论

学校读者我要写书评

暂无评论

Multi-stage hybrid multiobjective evolutionary algorithm for multi-type vehicle routing problem with simultaneous pickup and delivery and time windows

引用

Computers and Industrial Engineering 2025年 207卷

作者： Zhang, Wenqiang Li, Shun Deng, Miaolei Mu, Yashuang Li, Peng Ministry of Education China Henan Key Laboratory of Grain Storage Information Intelligent Perception and Decision Making Henan University of Technology China School of Artificial Intelligence and Big Data Henan University of Technology China College of Information Science and Engineering Henan University of Technology China Institute for Complexity Science Henan University of Technology China

The vehicle routing problem (VRP) is a fundamental and extensively studied problem in logistics and transportation. Despite its significance, real-world applications often require more complex variants to address specific operational constraints and objectives. To address these limitations, this paper introduces the multi-type vehicle routing problem with simultaneous pickup and delivery and time windows (MTVRPSPDTW). Although hybrid multiobjective algorithms have been successful in combination optimization problems in recent years, it is still challenging to improve the performance of the algorithms by combining them with different timing. We propose a multi-stage hybrid evolutionary multi-objective optimization with a multi-region sampling strategy (MS-HEMO-MRSS) to optimize both vehicle number and wait time of MTVRPSPDTW. The algorithm integrates a three-stage hybrid approach, combining a global search using the multi-region sampling strategy (MRSS) with a local search based on routing sequence differential evolution (RSDE). The initial stage employs MRSS to quickly position the population near the Pareto front from various directions. The second stage utilizes RSDE to accelerate convergence towards central and edge areas of the Pareto front. In the final stage, individuals on Pareto front are selected and RSDE is used again to guide them towards the edge regions to enhance distribution performance. Specialized encoding, decoding techniques, and genetic operators tailored for two vectors are introduced to optimize MTVRPSPDTW. Comparative experiments with traditional multiobjective evolutionary algorithms demonstrate significant convergence and notable distribution performances, highlighting the benefits of employing different optimization strategies at different evolutionary stages. © 2025 Elsevier Ltd

关键词： Pareto principle

来源：评论

学校读者我要写书评

暂无评论

On the Stability and Generalization of Triplet Learning

arXiv

引用

arXiv 2023年

作者： Chen, Jun Chen, Hong Jiang, Xue Gu, Bin Li, Weifu Gong, Tieliang Zheng, Feng College of Informatics Huazhong Agricultural University Wuhan430070 China College of Science Huazhong Agricultural University Wuhan430070 China Engineering Research Center of Intelligent Technology for Agriculture Ministry of Education Wuhan430070 China Key Laboratory of Smart Farming for Agricultural Animals Wuhan430070 China Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen518055 China Mohamed bin Zayed University of Artificial Intelligence Abu Dhabi United Arab Emirates School of Computer Science and Technology Xi'an Jiaotong University Xi’an710049 China Shaanxi Provincial Key Laboratory of Big Data Knowledge Engineering Ministry of Education Xi’an710049 China

Triplet learning, i.e. learning from triplet data, has attracted much attention in computer vision tasks with an extremely large number of categories, e.g., face recognition and person re-identification. Albeit with rapid progress in designing and applying triplet learning algorithms, there is a lacking study on the theoretical understanding of their generalization performance. To fill this gap, this paper investigates the generalization guarantees of triplet learning by leveraging the stability analysis. Specifically, we establish the first general high-probability generalization bound for the triplet learning algorithm satisfying the uniform stability, and then 1 obtain the excess risk bounds of the order (Equation presented). for both stochastic gradient descent (SGD) and regularized risk minimization (RRM), where 2n is approximately equal to the number of training samples. Moreover, an optimistic generalization bound in expectation as fast as O(n−1) is derived for RRM in a low noise case via the on-average stability analysis. Finally, our results are applied to triplet metric learning to characterize its theoretical underpinning. Copyright © 2023, The Authors. All rights reserved.

关键词： Learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Evaluation Methods of English Advanced Pronunciation Skills Based on Speech Recognition 4th

Evaluation Methods of English Advanced Pronunciation Skills ...

引用

4th International Conference on Intelligent, Interactive Systems and Applications, IISA2019

作者： Meng, Feng Jia, Yanming Wang, Wufeng Ning, Xu School of Computer Science and Technology Wuhan University of Technology Wuhan430070 China Center for Artificial Intelligence and Big Data Research Global Wisdom Inc. Beijing100085 China Zhongshan Institute of Advanced Engineering Technology of WUT Zhongshan528437 China Hubei Key Laboratory of Transportation Internet of Things Wuhan University of Technology Wuhan430070 China

ISBN: (纸本)9783030343866

The assessment of English advanced pronunciation skills evaluates the pronunciation fluency, employing techniques from various fields including artificial neural network, speech recognition and linguistics. The pronunciation fluency includes 4 main parts of pronunciation skills, including speech stress, pause, liaison and plosive. This paper mainly studies liaison and plosive. In the aspect of liaison recognition, GRU (Gated Recurrent Unit) are used to train the classification model. In the field of plosive recognition, words both with l plosive label or not are added to the process of language model construction;and acoustic features both with plosive pronunciation skill or not are used to the process of acoustic model construction, with these two parts, static decoding network WFST (Weight Finite-State Transducer) is constructed. The experimental results show that, for different types of liaison, the average accuracy of GRU models are up to 92.34%;for plosive model, the accuracy of the static decoding network model is 72.17%. © 2020, Springer Nature Switzerland AG.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

A robust multi-scale feature fusion model for low-quality multi-source data in underwater environments

引用

Computers and Electrical Engineering 2025年 126卷

作者： Yixiang Luo Ning Li Yuting Zhang Mengyun Liu Yun Peng Yuyan Luo Xiaoying Wang Institute of Artificial Intelligence Guangzhou University Guangzhou 510000 China Guangdong Key Laboratory of Blockchain Security Guangzhou University Guangzhou 510000 China School of Artificial intelligence and Big Data Henan University of Technology Zhengzhou 450001 China Third Affiliated Hospital of Sun Yat-sen University Guangzhou 510630 China

Efficient and accurate perception of complex underwater scenes is crucial for ensuring the success of subsequent tasks. Multi-source image fusion techniques offer an effective solution, however, the presence of complex factors such as feature distortion, imaging blur, and lighting variations in low-quality multi-source (sonar-optical) underwater images leads to significant degradation in fusion performance. To address this issue, we propose a novel underwater multi-source data fusion model, incorporating multi-scale features detection and fusion. First, we extract shallow and deep features from multi-source data to detect rich local texture features and global structural features. Then, the detailed features and semantic information in the fusion process were enhanced through the designed multi-scale feature fusion module, and the problems such as low saturation and partial feature loss in the fusion image reconstruction were alleviated. This provides accurate multi-source fusion capabilities for downstream tasks. Extensive experiments on the public dataset demonstrate that our fusion method significantly improves the performance of tasks by 0.74% and 3.34%, surpassing related state-of-the-art methods.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Reliability on Deep Learning Models: A Comprehensive Observation

Reliability on Deep Learning Models: A Comprehensive Observa...

引用

International Symposium on System and Software Reliability (ISSSR)

作者： Yuhong Zhang Chunjing Xiao School of Artificial Intelligence and Big Data Key Laboratory of Grain Information Processing and Control Ministry of Education Henan University of Technology Zhengzhou China Henan Key Laboratory of Big Data Analysis and Processing Henan University Kaifeng China

ISBN: (数字)9780738104973

ISBN: (纸本)9781665423236

This paper provides a comprehensive observation to examine the reliability of deep learning (DL) models. First, we will briefly introduce the essential background and kernel techniques in deep learning, such as downsampling and nonlinear discontinuity. Each of them may have some relation to the reliability of DL models. Then we discuss the inherent structural flaws of deep learning and the risk of unreliability that can result from it. Subsequently, we discuss various ways of generating adversarial samples that affect the DL model's reliability and corresponding preventive measures. Finally, we complete this observation by identifying current challenges and future trends for research.

关键词： Reliability Deep learning Biological neural networks Perturbation methods Neural networks Feature extraction data models

来源：评论

学校读者我要写书评

暂无评论

Low Rank Reduced Biquaternion Tensor Ring Decomposition and Tensor Completion

SSRN

引用

SSRN 2024年

作者： Luo, Hui Liu, Xin Liu, Wei Zhang, Yang School of Computer Science and Engineering Faculty of Innovation Engineering Macau University of Science and Technology Avenida Wai Long TaiPa 999078 China Macau Institute of Systems Engineering Faculty of Innovation Engineering Macau University of Science and Technology Avenida Wai Long TaiPa 999078 China School of Artificial Intelligence Sun Yat-sen University The Guangdong Key Laboratory of Big Data Analysis and Processing No.135 West Xingang Road Guangdong Provience Guangzhou510275 China Department of Mathematics University of Manitoba No.135 West Xingang Road WinnipegMBR3T 2N2 Canada

Reduced biquaternions (RBs), as an extension of complex numbers, have demonstrated exceptional capabilities in digital signal and image processing. They not only excel particularly in encoding color information but also feature multiplicative commutativity, which simplify the theoretical analysis of related algorithms. Concurrently, tensor completion methods based on tensor ring decomposition in the real domain have been proven effective in exploiting low-rank structures. Building upon these insights, this paper first introduces the Reduced Biquaternion Tensor Ring (RBTR) decomposition, along with a detailed exposition of the corresponding algorithm RBTR-SVD. Leveraging RBTR decomposition, we propose a novel low-rank tensor completion algorithm RBTR-TV integrating RBTR rank with Total Variation (TV) regularization to optimize the process. In color image completion tasks, our method exhibits remarkable stability and robustness, outperforming traditional methods, especially if there are a lot of missing data in the image. © 2024, The Authors. All rights reserved.

关键词： Tensors

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：