检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Chen, Xiaoyu Hu, Jiachen Yang, Lin F. Wang, Liwei Key Laboratory of Machine Perception MOE School of Artificial Intelligence Peking University China Electrical and Computer Engineering Department University of California Los Angeles United States International Center for Machine Learning Research Peking University China

Although model-based reinforcement learning (RL) approaches are considered more sample efficient, existing algorithms are usually relying on sophisticated planning algorithm to couple tightly with the model-learning procedure. Hence the learned models may lack the ability of being re-used with more specialized planners. In this paper we address this issue and provide approaches to learn an RL model efficiently without the guidance of a reward signal. In particular, we take a plug-in solver approach, where we focus on learning a model in the exploration phase and demand that any planning algorithm on the learned model can give a near-optimal policy. specifically, we focus on the linear mixture MDP setting, where the probability transition matrix is a (unknown) convex combination of a set of existingmodels. We show that, by establishing a novel exploration algorithm, the plug-in approach learns a model by taking O(d2H3/ϵ2) episodes with the environment and any ϵ-optimal planner on the model gives an O(ϵ)-optimal policy on the original model. This sample complexity matches our lower bound for non-plug-in approaches and is statistically optimal. We achieve this result by leveraging a careful maximum total-variance bound using Bernstein inequality and properties specified to linear mixture MDPs. Copyright © 2021, The Authors. All rights reserved.

关键词： Mixtures

来源：评论

学校读者我要写书评

暂无评论

A Prediction Approach Based on Long Short-Term Memory Networks for Dynamic Multiobjective Optimization

SSRN

引用

SSRN 2024年

作者： Xu, Biao Rang, Gejie Li, Wenji Fan, Zhun Gong, Dunwei Yang, Shengxiang He, Jie College of Engineering Shantou University Shantou515063 China College of Automation and Electronic Engineering Qingdao University of Science and Technology Qingdao266100 China Guangxi Key Laboratory of Machine Vision and Intelligent Control Wuzhou University Wuzhou543002 China School of Computer Science and Informatics De Montfort University LeicesterLE1 9BH United Kingdom School of Energy and Mechanical Engineering Nanchang Institute of Technology Nanchang330013 China Minjiang University China

Dynamic multiobjective optimization problems (DMOPs) have posed considerable hurdles for traditional evolutionary optimization techniques due to the perpetual variations in their Pareto-optimal sets (PSs) and fronts (PFs). Prediction-based methods have shown promise in swiftly tracking these alterations. However, the majority of current methods rely on linear models to anticipate changing PSs, which can be limiting. When PSs across different environments exhibit nonlinear patterns, accurately predicting the PS in a novel problem becomes challenging. To address this issue, this study introduces a dynamic multiobjective evolutionary optimization algorithm leveraging long short-term memory (LSTM) network predictions. In our approach, the PS of each problem is first divided into several clusters, and the center of each cluster can then be obtained. Thereafter, the cluster centers for the PSs of the problems with different environments form a number of time series. Subsequently, the LSTM network models are trained using the time seriesdata as training samples. Furthermore, the cluster centers of the new PS are predicted utilizing the trained models. LSTM networks excel at capturing historical information and modeling nonlinear relationships, ensuring high prediction accuracy. To enhance the algorithm’s performance, a novel population-generation strategy is employed, ensuring a well-converged and diverse initial population. The proposed algorithm is rigorously tested using benchmark functions, and the results demonstrate its effectiveness in handling DMOPs, outperforming state-of-the-art algorithms. © 2024, The Authors. All rights reserved.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Object Segmentation based on Anti-disturb Network in Ecological Surveillance

Object Segmentation based on Anti-disturb Network in Ecologi...

引用

第33届中国控制与决策会议

作者： Wan Dai Hongpeng Wang Chongshan Fan Yunwei Xin College of Computer Science Nankai University Tianjin Key Laboratory of Intelligent Robotics College of Artificial Intelligence Nankai University Software College Nankai University Research Laboratory of Machine Learning and Pervasive Computing Nankai University

Visual tele-observation is an effective way for ecological *** from pedestrian or rigid body detection,wildlife recognition in natural scenes face more complex problems,such as the similarity of background and foreground,wild background clutter,deformation,and other factors,which will lead to a reduction in recognition *** this paper,We propose a novel segmentation network that is named T-S network to segment background and *** network mainly consists of two parts:one is a tracking *** the context similarity of images in the data set,the purpose of tracking is to ensure that the position of the object is accurately known when the same species in the images are synthesized into a video with interval frames;the other is the saliency network,which is the basis of the most likely position of the object obtained from the tracking ***,the saliency network is used to complete the object *** verify the effectiveness of the network and consider the size of the data set,a lightweight CNN is used to identify the *** show that the accuracy of task recognition is improved by 1.89% by using the segmentation method proposed in this paper.

关键词： segmentation recognition tracking network saliency network

来源：评论

学校读者我要写书评

暂无评论

A Prediction Approach Based on Long Short-Term Memory Networks for Dynamic Multiobjective Optimization

SSRN

引用

SSRN 2024年

作者： Xu, Biao Rang, Gejie Li, Wenji Gong, Dunwei Fan, Zhun Yang, Shengxiang He, Jie College of Engineering Shantou University Shantou515063 China College of Automation and Electronic Engineering Qingdao University of Science and Technology Qingdao266100 China Guangxi Key Laboratory of Machine Vision and Intelligent Control Wuzhou University Wuzhou543002 China Shenzhen institute for Advanced Study UESTC Shenzhen518110 China School of Computer Science and Informatics De Montfort University LeicesterLE1 9BH United Kingdom Minjiang University China

Dynamic multiobjective optimization problems (DMOPs) present significant challenges to conventional evolutionary optimization methods because of the continuous changes in their Pareto-optimal sets (PSs) and fronts (PFs). Prediction-driven approaches have demonstrated potential in rapidly adapting to these changes. However, many existing methods depend on linear models to forecast the evolving PSs, which may be restrictive. To counteract this limitation, this research presents a novel dynamic multiobjective evolutionary optimization algorithm that incorporates predictions from long short-term memory (LSTM) networks. Initially, in our methodology, the PS for each problem is segmented into multiple clusters, and the centroid of each cluster is identified. These cluster centroids, representing the PSs across various environmental conditions, are then transformed into a series of time series data. The LSTM network models are subsequently trained on this time series data as input samples. Utilizing these refined models, the centroids of the evolving PSs are predicted with improved precision. Moreover, to enhance the performance of the algorithm, an innovative population-generation strategy is also introduced that guarantees a well-converged and diverse starting population. Our proposed algorithm undergoes rigorous testing using benchmark functions, and the outcomes validate its proficiency in tackling DMOPs, showing superior performance compared to existing state-of-the-art algorithms. © 2024, The Authors. All rights reserved.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Exploiting Completeness and Uncertainty of Pseudo Labels for Weakly Supervised Video Anomaly Detection

arXiv

引用

arXiv 2022年

作者： Zhang, Chen Li, Guorong Qi, Yuankai Wang, Shuhui Qing, Laiyun Huang, Qingming Yang, Ming-Hsuan State Key Laboratory of Information Security Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China School of Computer Science and Technology University of Chinese Academy of Sciences China Australian Institute for Machine Learning The University of Adelaide Australia Key Laboratory of Intelligent Information Processing Institute of Computing Technology CAS China Peng Cheng Laboratory China University of California Merced United States

Weakly supervised video anomaly detection aims to identify abnormal events in videos using only video-level labels. Recently, two-stage self-training methods have achieved significant improvements by self-generating pseudo labels and self-refining anomaly scores with these labels. As the pseudo labels play a crucial role, we propose an enhancement framework by exploiting completeness and uncertainty properties for effective self-training. Specifically, we first design a multi-head classification module (each head serves as a classifier) with a diversity loss to maximize the distribution differences of predicted pseudo labels across heads. This encourages the generated pseudo labels to cover as many abnormal events as possible. We then devise an iterative uncertainty pseudo label refinement strategy, which improves not only the initial pseudo labels but also the updated ones obtained by the desired classifier in the second stage. Extensive experimental results demonstrate the proposed method performs favorably against state-of-the-art approaches on the UCF-Crime, TAD, and XD-Violence benchmark datasets. Copyright © 2022, The Authors. All rights reserved.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

UNDERSTANDING DOMAIN RANDOMIZATION for SIM-TO-REAL TRANSFER

arXiv

引用

arXiv 2021年

作者： Chen, Xiaoyu Hu, Jiachen Jin, Chi Li, Lihong Wang, Liwei Key Laboratory of Machine Perception MOE School of Artificial Intelligence Peking University China Department of Electrical and Computer Engineering Princeton University United States Amazon United States International Center for Machine Learning Research Peking University China

Reinforcement learning encounters many challenges when applied directly in the real world. Sim-to-real transfer is widely used to transfer the knowledge learned from simulation to the real world. Domain randomization-one of the most popular algorithms for sim-to-real transfer-has been demonstrated to be effective in various tasks in robotics and autonomous driving. Despite its empirical successes, theoretical understanding on why this simple algorithm works is limited. In this paper, we propose a theoretical framework for sim-to-real transfers, in which the simulator is modeled as a set of MDPs with tunable parameters (corresponding to unknown physical parameters such as friction). We provide sharp bounds on the sim-to-real gap-the difference between the value of policy returned by domain randomization and the value of an optimal policy for the real world. We prove that sim-to-real transfer can succeed under mild conditionswithout any real-world training samples. Our theory also highlights the importance of using memory (i.e., history-dependent policies) in domain randomization. Our proof is based on novel techniques that reduce the problem of bounding the sim-to-real gap to the problem of designing efficient learning algorithms for infinite-horizon MDPs, which we believe are of independent interest. Copyright © 2021, The Authors. All rights reserved.

关键词： Random processes

来源：评论

学校读者我要写书评

暂无评论

High-speed and low-power molecular dynamics processing unit(MDPU)with ab initio accuracy

引用

npj Computational Materials 2024年第1期10卷 559-568页

作者： Pinghui Mo Yujia Zhang Zhuoying Zhao Hanhan Sun Junhua Li Dawei Guan Xi Ding Xin Zhang Bo Chen Mengchao Shi Duo Zhang Denghui Lu Yinan Wang Jianxing Huang Fei Liu Xinyu Li Mohan Chen Jun Cheng Bin Liang Weinan E Jiayu Dai Linfeng Zhang Han Wang Jie Liu College of Integrated Circuits Hunan UniversityChangsha410082P.R.China College of Electrical and Information Engineering Hunan UniversityChangsha410082P.R.China College of Computer National University of Defense TechnologyChangsha410073P.R.China Department of Physics National University of Defense TechnologyChangsha410073P.R.China DP Technology 100080 BeijingP.R.China AI for Science Institute 100080 BeijingP.R.China Academy for Advanced Interdisciplinary Studies Peking University100871 BeijingP.R.China HEDPS CAPTCollege of EngineeringPeking University100871 BeijingP.R.China School of Mathematical Science Peking University100871 BeijingP.R.China State Key Laboratory of Physical Chemistry of Solid Surfaces iChEMCollege of Chemistry and Chemical EngineeringXiamen UniversityXiamen361005P.R.China School of Integrated Circuits Peking University100871 BeijingP.R.China Center for Machine Learning Research Peking University100871 BeijingP.R.China Institute of Applied Physics and Computational Mathematics 100088 BeijingP.R.China Greater Bay Area Institute for Innovation Hunan UniversityGuangzhou511300P.R.China Department of Electrical and Computer Engineering University of WashingtonSeattleWA98195USA

Molecular dynamics(MD)is an indispensable atomistic-scale computational tool widely-used in various *** the past decades,nearly all ab initio MD and machine-learning MD have been based on the general-purpose central/graphics processing units(CPU/GPU),which are well-known to suffer from their intrinsic“memory wall”and“power wall”***,nowadays MD calculations with ab initio accuracy are extremely time-consuming and power-consuming,imposing serious restrictions on the MD simulation size and *** solve this problem,here we propose a special-purpose MD processing unit(MDPU),which could reduce MD time and power consumption by about 103 times(109 times)compared to state-of-the-art machine-learningMD(ab initio MD)based on CPU/GPU,while keeping ab initio *** significantly-enhanced performance,the proposed MDPU may pave a way for the accurate atomistic-scale analysis of large-size and/or longduration problems which were impossible/impractical to compute before.

关键词： consuming power MDP

来源：评论

学校读者我要写书评

暂无评论

Analysis of the Research Status of Information Technology Education Literature in China and Abroad

Analysis of the Research Status of Information Technology Ed...

引用

IEEE International Conference on High Performance Computing and Communications (HPCC)

作者： Yulin Zhao Kai Liu Junke Li School of Computer and Information Qiannan Normal University for Nationalities Duyun China Key Laboratory of Machine learning and Unstructured Data Processing of Qiannan Duyun China School of Information Engineering Suqian University Suqian Jiangsu China Key Laboratory of Complex Systems and Intelligent Optimization of Guizhou Duyun China Jiangsu Province Engineering Research Center of Smart Poultry Farming and Intelligent Equipment Jiangsu China

Information technology education contributes to the development of national education. Understanding the research status of information technology education at home and abroad is helpful to the implementation of education informatization. This paper takes the information technology education research literature collected by CNKI and the Web of Science database as the object and uses the bibliometric method to analyze the literature characteristics of information technology education at home and abroad. Specifically, the paper mainly carries out statistical analysis from the time distribution of literature, journal distribution, and hot keywords, and visually describes the literature features of the information technology education discipline, providing references for the research of information technology education discipline at home and abroad.

关键词： Statistical analysis Databases Education Bibliometrics Educational technology Market research Internet

来源：评论

学校读者我要写书评

暂无评论

B2e-Cdg: Artifact Removal for Enhanced Optical Coherence Tomography Angiography Using Conditional Guidance Diffusion

SSRN

引用

SSRN 2024年

作者： Xu, Jing Fu, Suzhong Xing, Jiwei Xue, Linyan Wang, Cheng Tuchin, Valery V. Zhao, Qingliang State Key Laboratory of Molecular Vaccinology and Molecular Diagnostics Organ Transplantation Institute China Institute of Artificial Intelligence Xiamen University Xiamen361102 China State Key Laboratory of Vaccines for Infectious Diseases Xiang An Biomedicine Laboratory School of Public Health Xiamen University Xiamen361102 China Shenzhen518172 China College of Quality and Technical Supervision Hebei University Baoding071002 China School of Informatics Computer Science and Technology Department Xiamen University Xiamen361005 China Institute of Physics and Science Medical Center Saratov State University Saratov410012 Russia Laboratory of Laser Molecular Imaging and Machine Learning Tomsk State University Tomsk634050 Russia Institute of Precision Mechanics and Control FRS "Saratov Scientific Centre of the RAS" Saratov410028 Russia China

Optical Coherence Tomography Angiography (OCTA) is a revolutionary technology widely used to diagnose and manage fundus, skin and cardiovascular diseases. However, unavoidable movements, such as breathing, cause motion artifacts in images, which can significantly degrade image quality, obscure critical vascular details, and reduce the diagnostic reliability of the modality. Although recent advances in learning-based image inpainting methods for OCTA enface images have made notable progress in artifact removal, these methods still require the collection of large amounts of accurately labeled data and the generation of pseudo stripes to create paired training sets. Furthermore, the abundant structural information and flow intensity signals present in OCTA B-scans are nonnegligible. Hence we proposed B-scans to Enface Conditional Diffusion Guidance (B2E-CDG) for translation from signal-void B-scans to available B-scans. We introduce the normal B-scans in a connection manner and the specified reference B-scans in a gradient-based manner as style feature guidance into the diffusion model. Conditional guidance facilitates a more controllable and precise generation process for B-scans flow signal recovery. The requirement for labeled image collection and pseudo stripes is obviated, as the repetitive scanning nature of OCTA naturally results in paired datasets comprising signal-void and normal B-scans. The experimental results demonstrated that B2E-CDG effectively removes artifacts and restores vascular and structural information obscured by artifacts. This work shows superior vascular recovery and artifact removal capabilities in metrics and downstream tasks, enhancing the usability of OCTA in diagnostics. © 2024, The Authors. All rights reserved.

关键词： Optical coherence tomography

来源：评论

学校读者我要写书评

暂无评论

Weakly-Supervised Depression Detection in Speech Through Self-learning Based Label Correction

IEEE Transactions on Audio, Speech and Language Processing

引用

IEEE Transactions on Audio, Speech and Language Processing 2025年 33卷 748-758页

作者： Yanfei Sun Yuanyuan Zhou Xinzhou Xu Jin Qi Feiyi Xu Zhao Ren Björn W. Schuller School of Internet of Things Nanjing University of Posts and Telecommunications Nanjing China Wuxi University Wuxi China School of Computer Science Nanjing University of Posts and Telecommunications Nanjing China Signal Processing and Speech Communication Laboratory Graz University of Technology Graz Austria Key Laboratory of Modern Acoustics MOE Nanjing University Nanjing China Cognitive Systems Lab University of Bremen Bremen Germany Chair of Health Informatics Technische Universität München (TUM) München Germany Munich Data Science Institute Munich Germany Munich Center for Machine Learning Munich Germany GLAM – the Group on Language Audio & Music Imperial College London London U.K.

Automated Depression Detection (ADD) in speech aims to automatically estimate one's depressive attributes through artificial intelligence tools towards spoken signals. Nevertheless, existing speech-based ADD works fail to sufficiently consider weakly-supervised cases with inaccurate labels, which may typically appear in intelligent mental health. In this regard, we propose the Self-learning-based Label Correction (SLLC) approach for weakly-supervised depression detection in speech. The proposed approach employs a self-learning manner connecting a label correction module and a depression detection module. Within the approach, the label correction module fuses likelihood-ratio-based and prototype-based label correction strategies in order to effectively correct the inaccurate labels, while the depression detection module aims at detecting depressed samples through a 1D convolutional recurrent neural network with multiple types of losses. The experimental results on two depression detection corpora show that our proposed SLLC approach performs better compared with existing state-of-the-art speech-based depression detection approaches, in the case of weak supervision with inaccurate labels for depression detection in speech.

关键词： Depression Feature extraction Training Speech processing Prototypes Noise measurement Telecommunications Sun Internet of Things Weak supervision

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：