检索结果-内蒙古大学图书馆

IEEE Transactions on Artificial Intelligence 2024年第7期5卷 3607-3623页

作者： Mandal, Lakshmi Diddigi, Raghuram Bharadwaj Bhatnagar, Shalabh Indian Institute of Science Department of Computer Science and Automation Karnataka Bangalore560012 India International Institute of Information Technology Karnataka Bangalore560100 India

Reinforcement learning (RL) algorithms combined with deep learning architectures have achieved tremendous success in many practical applications. However, the policies obtained by many deep reinforcement learning (DRL) algorithms are seen to suffer from high variance making them less useful in safety-critical applications. In general, it is desirable to have algorithms that give a low iterate-variance while providing a high long-term reward. In this work, we consider the actor-critic (AC) paradigm, where the critic is responsible for evaluating the policy while the feedback from the critic is used by the actor for updating the policy. The updates of both the critic and the actor in the standard AC procedure are run concurrently until convergence. It has been previously observed that updating the actor once after every L>1 steps of the critic reduces the iterate variance. In this article, we address the question of what optimal L-value to use in the recursions and propose a data-driven L-update rule that runs concurrently with the AC algorithm with the objective being to minimize the variance of the infinite horizon discounted reward. This update is based on a random search (discrete) parameter optimization procedure that incorporates smoothed functional (SF) estimates. We prove the convergence of our proposed multitimescale scheme to the optimal L and optimal policy pair. Subsequently, through numerical evaluations on benchmark RL tasks, we demonstrate the advantages of our proposed algorithm over multiple state-of-the-art algorithms in the literature. © 2020 IEEE.

关键词： Optimization

来源：评论

学校读者我要写书评

暂无评论

Quality-Relevant Modeling and Monitoring of Industrial Cyber-Physical Systems: The Semi-Supervised Dynamic Latent Variable Models

IEEE Transactions on Industrial Cyber-Physical Systems

引用

IEEE Transactions on Industrial Cyber-Physical Systems 2025年 3卷 39-47页

作者： Zhou, Le Wang, Yaoxin Wu, Yuanqing He, Shenghuang Song, Zhihuan Zhejiang University of Science and Technology Department of Automation and Electrical Engineering Hangzhou310023 China Guangdong University of Technology Department of Automation Guangzhou510006 China Dongguan University of Technology Department of Computer Science and Technology Dongguan523808 China Ningbo Industrial Internet Institute Ningbo315012 China Zhejiang University Department of Control Science and Engineering Hangzhou310027 China

In modern industrial cyber-physical systems, a mass of process variables has been obtained by the high-sampling online sensors. Meanwhile, the key quality indexes are usually obtained infrequently from the laboratory. Hence, these quality variables are with low sampling rate. To avail of the complete process and quality variables with various sampling rates in the dynamic processes, a set of semi-supervised dynamic latent variable models are proposed for dynamic modeling and quality-relevant monitoring. The proposed models have built a unified structure to consider both the auto-correlations and cross-correlations between the process and quality variables with unbalanced sampling sizes. Hence, the feature extraction of the time series data is dynamically adjusted under the guidance of the quality variables. Then, the quality-relevant monitoring schemes are proposed, which is validated by a numerical case and an actual wastewater treatment process. © 2024 IEEE. All rights reserved.

关键词： Indexes Probabilistic logic Kalman filters Noise Cyber-physical systems Analytical models Parameter estimation Time series analysis Process monitoring Numerical models

来源：评论

学校读者我要写书评

暂无评论

Unpredictability of Digital Twin for Connected Vehicles

引用

China Communications 2023年第2期20卷 26-45页

作者： Xiaoxu Wang Zeyin Huang Songmiao Zheng Rong Yu Miao Pan School of Automation Guangdong University of TechnologyGuangzhou 510006China Electrical and Computer Engineering Department University of HoustonHoustonTX 77004USA

Digital twin is an essential enabling technology for 6G connected *** highfidelity mobility simulation,digital twin is expected to make accurate prediction about the vehicle trajectory,and then support intelligent applications such as safety monitoring and self-driving for connected ***,it is observed that even if a digital twin model is perfectly derived,it might still fail to predict the trajectory due to tiny measurement noise or delay in the initial vehicle *** paper aims at investigating the sources of unpredictability of digital *** the car-following behaviors in connected vehicles for case *** theoretical analysis and experimental results indicate that the predictability of digital twin naturally depends on its system *** a system enters a complex pattern,its longterm states are ***,our study discloses that the complexity is determined,on the one hand,by the intrinsic factors of the target physical system such as the driver’s response sensitivity and delay,and on the other hand,by the crucial parameters of the digital twin system such as the sampling interval and twining latency.

关键词： digital twin connected vehicles unpredictability complexity

来源：评论

学校读者我要写书评

暂无评论

An Optimal Weighted Ensemble of 3D CNNs for Early Diagnosis of Alzheimer’s Disease

引用

SN computer Science 2024年第2期5卷 252页

作者： Dharwada, Sriram Tembhurne, Jitendra Diwan, Tausif Department of Computer Science and Automation Indian Institute of Science Bangalore India Department of Computer Science and Engineering Indian Institute of Information Technology Maharashtra Nagpur 441108 India

Alzheimer disease (AD) is a chronic neurological disorder in which the loss of brain cells causes dementia. Early and accurate diagnosis of AD will lead to better treatment of the disease before irreversible brain damage has been occurred. This paper proposes the classification of Alzheimer's disease using 3D structural Magnetic Resonance Imaging (sMRI) images through 3D convolutional neural networks (CNNs). Most existing methods utilizing 3D subject-level CNNs for Alzheimer's disease classification design a single model which relies on a very large training dataset for improved generalization. Herein, we address this issue through 3D transfer learning which makes use of knowledge gained from a pre-trained task. We train 3D versions of five classical 2D image classification architectures—ResNet, ResNeXt, SE-ResNet, SE-ResNeXt, and SE-Net—by initializing each model with pre-trained weights from their 2D counterparts, and combine their predictions through a weighted average method. The weights assigned to each model of the ensemble are optimized to achieve a performance better than any single 3D CNN model. With a relatively smaller training dataset, the proposed model obtains 97.27%, 82.33%, 90.41%, 84.22%, 84.26%, and 77.1% accuracies for the Alzheimer’s disease (AD) versus cognitively normal (CN), early mild cognitive impairment (EMCI) versus CN, late mild cognitive impairment (LMCI) versus CN, EMCI versus AD, LMCI versus AD, and EMCI versus LMCI classification tasks, outperforming current state-of-the-art methods, and indicating the effectiveness of our proposed model. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd 2024.

关键词： Alzheimer’s disease Convolutional neural networks Ensemble learning Transfer learning

来源：评论

学校读者我要写书评

暂无评论

Optimal Production Capacity Matching for Blockchain-Enabled Manufacturing Collaboration With the Iterative Double Auction Method

引用

IEEE/CAA Journal of Automatica Sinica 2025年第3期12卷 550-562页

作者： Ying Chen Feilong Lin Zhongyu Chen Changbing Tang Cailian Chen the Key Laboratory of Intelligent Education Technology and Application of Zhejiang Province Zhejiang Normal University the School of Computer Science and Technology Zhejiang Normal University IEEE the College of Physics and Electronic Information Engineering Zhejiang Normal University the Department of Automation Shanghai Jiao Tong University

The increased demand for personalized customization calls for new production modes to enhance collaborations among a wide range of manufacturing practitioners who unnecessarily trust each other. In this article, a blockchain-enabled manufacturing collaboration framework is proposed, with a focus on the production capacity matching problem for blockchainbased peer-to-peer(P2P) collaboration. First, a digital model of production capacity description is built for trustworthy and transparent sharing over the blockchain. Second, an optimization problem is formulated for P2P production capacity matching with objectives to maximize both social welfare and individual benefits of all participants. Third, a feasible solution based on an iterative double auction mechanism is designed to determine the optimal price and quantity for production capacity matching with a lack of personal information. It facilitates automation of the matching process while protecting users' privacy via blockchainbased smart contracts. Finally, simulation results from the Hyperledger Fabric-based prototype show that the proposed approach increases social welfare by 1.4% compared to the Bayesian game-based approach, makes all participants profitable,and achieves 90% fairness of enterprises.

关键词： Blockchain iterative double auction manufacturing collaboration production capacity matching

来源：评论

学校读者我要写书评

暂无评论

Constrained Multi-Objective Optimization With Deep Reinforcement Learning Assisted Operator Selection

引用

IEEE/CAA Journal of Automatica Sinica 2024年第4期11卷 919-931页

作者： Fei Ming Wenyin Gong Ling Wang Yaochu Jin the School of Computer Science China University of GeosciencesWuhan 430074China the Department of Automation Tsinghua UniversityBeijing 100084China the Faculty of Technology Bielefeld UniversityNorth Rhine-Westphalia33619 BielefeldGermany

Solving constrained multi-objective optimization problems with evolutionary algorithms has attracted considerable *** constrained multi-objective optimization evolutionary algorithms(CMOEAs)have been developed with the use of different algorithmic strategies,evolutionary operators,and constraint-handling *** performance of CMOEAs may be heavily dependent on the operators used,however,it is usually difficult to select suitable operators for the problem at ***,improving operator selection is promising and necessary for *** work proposes an online operator selection framework assisted by Deep Reinforcement *** dynamics of the population,including convergence,diversity,and feasibility,are regarded as the state;the candidate operators are considered as actions;and the improvement of the population state is treated as the *** using a Q-network to learn a policy to estimate the Q-values of all actions,the proposed approach can adaptively select an operator that maximizes the improvement of the population according to the current state and thereby improve the algorithmic *** framework is embedded into four popular CMOEAs and assessed on 42 benchmark *** experimental results reveal that the proposed Deep Reinforcement Learning-assisted operator selection significantly improves the performance of these CMOEAs and the resulting algorithm obtains better versatility compared to nine state-of-the-art CMOEAs.

关键词： Constrained multi-objective optimization deep Qlearning deep reinforcement learning(DRL) evolutionary algorithms evolutionary operator selection

来源：评论

学校读者我要写书评

暂无评论

Real-Time Sketching of Harshly Lit Driving Environment Perception by Neuromorphic Sensing

IEEE Transactions on Intelligent Vehicles

引用

IEEE Transactions on Intelligent Vehicles 2024年 1-15页

作者： Han, Yuqi Suo, Jinli Zhang, Bo Zhao, Jiachen Cheng, Yuxiao Dai, Qionghai School of Computer Science and Technology Dalian University of Technology Liaoning Province China Department of Automation Tsinghua University Beijing China Beijing National Research Center for Information Science and Technology Tsinghua University Beijing China

Visual sensors are indispensable for automatic vehicles, to achieve comprehensive environmental perception for navigation, but their deteriorated performance in harsh illuminations largely sets back the practical use of autonomous driving technologies. A promising solution is to use a bio-inspired event sensor that asynchronously records the intensity changes with high sensitivity, fast response, and large dynamic range, which assists situational awareness of moving vehicles in harshly lit scenarios. However, the sensing of event sensors comes with heavy noise and sparse signals, due to either severe photon starvation or limited acquisition bandwidth. In this paper, we propose an approach for real-time sketching of the harshly lit driving environment (RIDE), to outline the driving surroundings from noisy sporadic measurements. We address confronted challenges as follows: (i) map the raw event signals into a low dimensional space and cluster the features to depict the spatial-temporal correlation within raw events;(ii) design a general inference network to construct continuous motion fields of the scene from the encoded features of noisy sporadic raw measurements;(iii) construct the pseudo-ground-truth via the unsupervised motion compensation as the label of the above network learning, achieving real-time inference. Our approach is experimentally validated on real traffic data and displays high-fidelity perception capability for extremely dark scenes and scenarios with high dynamic range. Also, we investigate RIDE's effectiveness in the downstream task—detection of traffic participants. In a nutshell, the proposed RIDE provides high-fidelity sensing of harshly lit environments and lays the foundation for the all-day visual navigation of autonomous vehicles. IEEE

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning with Quasi-Hyperbolic Discounting 63

Reinforcement Learning with Quasi-Hyperbolic Discounting

引用

63rd IEEE Conference on Decision and Control, CDC 2024

作者： Eshwar, S.R. Motwani, Mayank Roy, Nibedita Thoppe, Gugan Indian Institute of Science Department of Computer Science and Automation Bengaluru India Indian Institute of Technology Department of Computer Science and Engineering Bombay India

ISBN: (纸本)9798350316339

Reinforcement learning has traditionally been studied with exponential discounting or the average reward setup, mainly due to their mathematical tractability. However, such frameworks fall short of accurately capturing human behavior, which has a bias towards immediate gratification. Quasi-Hyperbolic (Q H) discounting is a simple alternative for modeling this bias. Unlike in traditional discounting, though, the optimal QH-policy, starting from some time t1, can be different to the one starting from t2. Hence, the future self of an agent, if it is naive or impatient, can deviate from the policy that is optimal at the start, leading to sub-optimal overall returns. To prevent this behavior, an alternative is to work with a policy anchored in a Markov Perfect Equilibrium (MPE). In this work, we propose the first model-free algorithm for finding an MPE. Using a two-timescale analysis, we show that, if our algorithm converges, then the limit must be an MPE. We also validate this claim numerically for the standard inventory system with stochastic demands. Our work significantly advances the practical application of reinforcement learning. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning with Quasi-Hyperbolic Discounting (Extended Abstract) 10

Reinforcement Learning with Quasi-Hyperbolic Discounting (Ex...

引用

10th Indian Control Conference, ICC 2024

作者： Eshwar, S.R. Motwani, Mayank Roy, Nibedita Thoppe, Gugan Department of Computer Science and Automation Indian Institute of Science Bengaluru India Department of Computer Science and Engineering Indian Institute of Technology Bombay India

ISBN: (纸本)9798331517212

Human decision-making is better modeled via discounting schemes such as quasi-hyperbolic and hyperbolic than classical ones such as exponential or average reward. In [3], we initiated the study of reinforcement learning with Quasi-Hyperbolic (QH) discounting and proposed the first model-free algorithm for finding a type of policy called the Markov perfect equilibrium. This paper provides a summary of that algorithm and our other findings on its convergence. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Geometric prior guided hybrid deep neural network for facial beauty analysis

引用

CAAI Transactions on Intelligence technology 2024年第2期9卷 467-480页

作者： Tianhao Peng Mu Li Fangmei Chen Yong Xu David Zhang The School of Computer Science and Technology Guizhou UniversityGuiyangChina Department of Automation Moutai InstituteRenhuaiGuizhouChina The School of Computer Science and Technology Harbin Institute of TechnologyShenzhenShenzhenChina The Information and Communication Engineering Department Dalian Minzu UniversityDalianChina The School of Data Science The Chinese University of Hong KongShenzhenShenzhenChina

Facial beauty analysis is an important topic in human *** may be used as a guidance for face beautification applications such as cosmetic *** neural networks(DNNs)have recently been adopted for facial beauty analysis and have achieved remarkable ***,most existing DNN-based models regard facial beauty analysis as a normal classification *** ignore important prior knowledge in traditional machine learning models which illustrate the significant contribution of the geometric features in facial beauty *** be specific,landmarks of the whole face and facial organs are introduced to extract geometric features to make the *** by this,we introduce a novel dual-branch network for facial beauty analysis:one branch takes the Swin Transformer as the backbone to model the full face and global patterns,and another branch focuses on the masked facial organs with the residual network to model the local patterns of certain facial ***,the designed multi-scale feature fusion module can further facilitate our network to learn complementary semantic information between the two *** model optimisation,we propose a hybrid loss function,where especially geometric regulation is introduced by regressing the facial landmarks and it can force the extracted features to convey facial geometric *** performed on the SCUT-FBP5500 dataset and the SCUT-FBP dataset demonstrate that our model outperforms the state-of-the-art convolutional neural networks models,which proves the effectiveness of the proposed geometric regularisation and dual-branch structure with the hybrid *** the best of our knowledge,this is the first study to introduce a Vision Transformer into the facial beauty analysis task.

关键词： deep neural networks face analysis face biometrics image analysis

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：