检索结果-内蒙古大学图书馆

Heavy-Ball Momentum Accelerated Actor-Critic With Function Approximation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Dong, Yanjie Zhang, Haijun Wang, Gang Cui, Shisheng Hu, Xiping The Artificial Intelligence Research Institute Guangdong-Hong Kong Macao Joint Laboratory for Emotional Intelligence and Pervasive Computing Shenzhen MSU-BIT University Shenzhen518172 China The Beijing Engineering and Technology Research Center for Convergence Networks and Ubiquitous Services University of Science and Technology Beijing Beijing China The State Key Laboratory of Autonomous Intelligent Unmanned Systems Beijing Institute of Technology Beijing100081 China

By using an parametric value function to replace the Monte-Carlo rollouts for value estimation, the actor-critic (AC) algorithms can reduce the variance of stochastic policy gradient so that to improve the convergence rate. While existing works mainly focus on analyzing convergence rate of AC algorithms under Markovian noise, the impacts of momentum on AC algorithms remain largely unexplored. In this work, we first propose a heavy-ball momentum based advantage actor-critic (HB-A2C) algorithm by integrating the heavy-ball momentum into the critic recursion that is parameterized by a linear function. When the sample trajectory follows a Markov decision process, we quantitatively certify the acceleration capability of the proposed HB-A2C algorithm. Our theoretical results demonstrate that the proposed HB-A2C finds an ǫ-approximate stationary point with O(ǫ−2) iterations for reinforcement learning tasks with Markovian noise. Moreover, we also reveal the dependence of learning rates on the length of the sample trajectory. By carefully selecting the momentum factor of the critic recursion, the proposed HB-A2C can balance the errors introduced by the initialization and the stoschastic approximation. © 2024, CC BY.

关键词： Stochastic systems

Collaborative Optimization of Cyber Physical Social systems for Urban Transportation Based on Knowledge Automation

学校读者我要写书评

暂无评论

IFAC-PapersOnLine 2020年第5期53卷 572-577页

The integration of signals from physical, social and cyber spaces, known as Cyber-Physical-Social systems (CPSS), is a new research paradigm for urban transportation, where the traffic control and management (C&M) is collaborative optimized among the three sub-systems. Though some technologies and optimization methods have been studied since its proposition, there is a lack of a systemic architecture as well as an overall implementation about how to efficiently exploit the social signals. For this reason, this paper proposes a general framework of CPSS for urban transportation and presents a feasible solution for traffic optimization based on knowledge automation. The specific implementation includes basic modeling of CPSS, knowledge evolution and reasoning, and collaborative optimization of C&P strategies. As a remarkable highlight, the influence of both individual activities and social learning is concerned during knowledge evolution and reasoning part. A case study from the application in the city of Dongguan is also given to validate our proposed framework and methods, showing that they can efficiently improve the average speed of the actual transportation.

关键词： CPSS knowledge automation urban transportation Collaborative Optimization

TiCoSS: Tightening the Coupling between Semantic Segmentation and Stereo Matching within A Joint Learning Framework

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tang, Guanfeng Wu, Zhiyuan Li, Jiahang Zhong, Ping Chen, Xieyuanli Lu, Huimin Fan, Rui The College of Electronics & Information Engineering Shanghai Research Institute for Intelligent Autonomous Systems Shanghai Institute of Intelligent Science and Technology Tongji University The State Key Laboratory of Intelligent Autonomous Systems Frontiers Science Center for Intelligent Autonomous Systems Tongji University Shanghai201804 China The Department of Engineering King's College London LondonWC2R 2LS United Kingdom The Department of Computer Science and Technology Central South University China The College of Intelligence Science and Technology National University of Defense Technology China

Semantic segmentation and stereo matching, respectively analogous to the ventral and dorsal streams in our human brain, are two key components of autonomous driving perception systems. Addressing these two tasks with separate networks is no longer the mainstream direction in developing computer vision algorithms, particularly with the recent advances in large vision models and embodied artificial intelligence. The trend is shifting towards combining them within a joint learning framework, especially emphasizing feature sharing between the two tasks. The major contributions of this study lie in comprehensively tightening the coupling between semantic segmentation and stereo matching. Specifically, this study introduces three key novelties: (1) a tightly coupled, gated feature fusion strategy, (2) a hierarchical deep supervision strategy, and (3) a coupling tightening loss function. All these technical contributions help tightly couple semantic segmentation and stereo matching tasks, collectively resulting in TiCoSS, a state-of-the-art joint learning framework. Through extensive experiments on the KITTI and vKITTI2 datasets, along with both qualitative and quantitative analyses, we validate the effectiveness of our developed strategies and loss function. Our approach demonstrates superior performance compared to prior arts, with a notable increse in mIoU by over 9%. Our source code will be publicly available at ***/TiCoSS upon publication. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantic Segmentation

The Winning Solution to the iFLYTEK Challenge 2021 Cultivated Land Extraction from High-Resolution Remote Sensing Image

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhao, Zhen Liu, Yuqiu Zhang, Gang Tang, Liang Hu, Xiaolin School of Technology Beijing Forestry University Beijing100091 China Department of Computer Science and Technology State Key Laboratory of Intelligent Technology and Systems Beijing National Research Center for Information Science and Technology Tsinghua University Beijing100084 China Tsinghua University Beijing100084 China Beijing100010 China

Extracting cultivated land accurately from high-resolution remote images is a basic task for precision agriculture. This report introduces our solution to the iFLYTEK challenge 2021 cultivated land extraction from high-resolution remote sensing image. The challenge requires segmenting cultivated land objects in very high-resolution multispectral remote sensing images. We established a highly effective and efficient pipeline to solve this problem. We first divided the original images into small tiles and separately performed instance segmentation on each tile. We explored several instance segmentation algorithms that work well on natural images and developed a set of effective methods that are applicable to remote sensing images. Then we merged the prediction results of all small tiles into seamless, continuous segmentation results through our proposed overlap-tile fusion strategy. We achieved the first place among 486 teams in the challenge. Copyright © 2022, The Authors. All rights reserved.

关键词：

Reports of the AAAI 2008 Spring Symposia

学校读者我要写书评

暂无评论

Reports of the AAAI 2008 Spring Symposia

作者： Balduccini, Marcello Baral, Chitta Brodaric, Boyan Colton, Simon Fox, Peter Gutelius, David Hinkelman, Knut Horswill, Ian Huberman, Bernardo Hudlicka, Eva Lerman, Kristina Lisetti, Christine McGuinness, Deborah Maher, Mary Lou Musen, Mark A. Sahami, Mehran Sleeman, Derek Thönssen, Barbara Velasquez, Juan Ventura, Dan Eastman Kodak Company Arizona State University Tempe AZ United States Geological Survey of Canada Department of Computing Imperial College London United Kingdom National Center for Atmospheric Research Artificial Intelligence Center SRI International Menlo Park CA United States University of Applied Sciences Zurich Switzerland Department of Computer Science Northwestern University Evanston IL United States Social Computing Lab. Hewlett-Packard Laboratories Psychometric Associates USC Information Sciences Institute University of Southern California Los Angeles CA United States Computer Science Department University of Southern California Los Angeles CA United States School of Computing and Information Sciences Florida International University Miami FL United States Rensselaer Polytechnic Institute Information and Intelligent Systems Division National Science Foundation University of Sydney Sydney NSW Australia Faculty of Architecture Design and Planning University of Sydney Sydney NSW Australia Department of Medicine Biomedica Informatics Research Department Stanford University Stanford CA United States Computer Science Department Stanford University Stanford CA United States Department of Computing Science University of Aberdeen Aberdeen United Kingdom MIT CSAIL Department of Computer Science Brigham Young University Provo UT United States

The Association for the Advancement of Artificial Intelligence (AAAI) was pleased to present the AAAI 2008 Spring Symposium Series, held Wednesday through Friday, March 26-28, 2008, at Stanford University, California. The eight symposia were titled (1) AI Meets Business Rules and Process Management, (2) architectures for intelligent Theory-Based Agents, (3) Creative intelligent systems, (4) Emotion, Personality, and Social Behavior, (5) Semantic Scientific Knowledge Integration, (6) Social Information Processing, (7) Symbiotic Relationships between Semantic Web and Knowledge Engineering, (8) Using Al to Motivate Greater Participation in computer Science. The goal of the AI Meets Business Rules and Process Management AAAI symposium was to investigate the various approaches and standards to represent business rules, business process management, and the semantic web with respect to expressiveness and reasoning capabilities. The focus of the architectures ror intelligent Theory-Based Agents AAAI symposium was the definition or architectures for intelligent theory-based agents, comprising languages, knowledge representation methodologies, reasoning algorithms, and control loops. The Creative intelligent systems symposium included five major discussion sessions and a general poster session (in which all contributing papers were presented). The purpose of this symposium was to explore the synergies between creative cognition and intelligent systems. The goal of the Emotion, Personality, and Social Behavior symposium was to examine fundamental issues in affect and personality in both biological and artificial agents, focusing on the roles of these factors in mediating social behavior. The Semantic Scientific Knowledge Integration symposium brought together the semantic technologies community with the scientific information technology community in an effort to build the general semantic science information community. The Social Information Processing symposium's goal was to

关键词： Knowledge representation

Towards High-accuracy and Low-latency Spiking Neural Networks with Two-Stage Optimization

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wang, Ziming Zhang, Yuhao Lian, Shuang Cui, Xiaoxin Yan, Rui Tang, Huajin College of Computer Science and Technology Zhejiang University Hangzhou310027 China Research Center for Intelligent Computing Hardware Zhejiang Lab Hangzhou311100 China School of Integrated Circuits Peking University Beijing100871 China College of Computer Science Zhejiang University of Technology Hangzhou310014 China State Key Laboratory of Brain-Machine Intelligence Zhejiang University Hangzhou310027 China

Spiking neural networks (SNNs) operating with asynchronous discrete events show higher energy efficiency with sparse computation. A popular approach for implementing deep SNNs is ANN-SNN conversion combining both efficient training of ANNs and efficient inference of SNNs. However, the accuracy loss is usually non-negligible, especially under few time steps, which restricts the applications of SNN on latencysensitive edge devices greatly. In this paper, we first identify that such performance degradation stems from the misrepresentation of the negative or overflow residual membrane potential in SNNs. Inspired by this, we decompose the conversion error into three parts: quantization error, clipping error, and residual membrane potential representation error. With such insights, we propose a two-stage conversion algorithm to minimize those errors respectively. Besides, We show each stage achieves significant performance gains in a complementary manner. By evaluating on challenging datasets including CIFAR-10, CIFAR- 100 and ImageNet, the proposed method demonstrates the stateof- the-art performance in terms of accuracy, latency and energy preservation. Furthermore, our method is evaluated using a more challenging object detection task, revealing notable gains in regression performance under ultra-low latency when compared to existing spike-based detection algorithms. Codes are available at https://***/Windere/snn-cvt-dual-phase. Copyright © 2022, The Authors. All rights reserved.

关键词： Object detection

In-Situ 3D Printing Monitoring in Dynamic Environments via Self-Supervised Deep Neural Network Adaptation

学校读者我要写书评

暂无评论

In-Situ 3D Printing Monitoring in Dynamic Environments via S...

IEEE International Conference on Automation Science and Engineering (CASE)

作者： Fang Wang Gang Xiong Qihang Fang Zhen Shen Di Wang Li Wan Xisong Dong Fei-Yue Wang State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing China School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China Beijing Engineering Research Center of Intelligent Systems and Technology Institute of Automation Chinese Academy of Sciences Beijing China Guangdong Engineering Research Center of 3D Printing and Intelligent Manufacturing Cloud Computing Center Chinese Academy of Sciences Dongguan China School of Mechanical and Automotive Engineering South China University of Technology Guangzhou China Beijing Ten Dimensions Technology Co. Ltd. Beijing China State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China

ISBN: (数字)9798350358513

ISBN: (纸本)9798350358520

3D printing technology has rapidly advanced, yet the variability in deployment environments poses significant challenges for monitoring systems, because the dynamic deployment environments such as changes in lighting and monitoring camera status, can severely impact the effectiveness and robustness of monitoring systems. To tackle this problem in monitoring system, this paper introduces a novel approach for deep neural network (DNN) adaptation to the dynamic environments through self-supervised learning and applies it to during in-situ monitoring. Specifically, we introduce a self-supervised learning strategy that leverages the auxiliary reconstruction task during in-situ monitoring, subsequently applying self-supervised fine-tuning to classification tasks with a new imbalanced-aware classification loss. Our methodology was rigorously evaluated using a real-world dataset for 3D printing defect detection. The experimental outcomes affirm the robustness of our approach, showcasing a higher defect detection accuracy rate than baselines. This substantially mitigates the adverse effects associated with printing defects, thereby increasing the reliability and quality of 3D printing processes.

关键词： Adaptation models Lighting Artificial neural networks Self-supervised learning Three-dimensional printing Feature extraction Robustness Monitoring Image reconstruction Defect detection

Network slicing for service-oriented networks under resource constraints

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Zhang, Nan Liu, Ya-Feng Farmanbar, Hamid Chang, Tsung-Hui Hong, Mingyi Luo, Zhi-Quan School of Mathematical Sciences Peking University State Key Laboratory of Scientific and Engineering Computing Institute of Computational Mathematics and Scientific/Engineering Computing Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing China Huawei Canada Research Center Ottawa Canada Department of Electrical and Computer Engineering University of Minnesota United States Shenzhen Research Institute of Big Data Chinese University of Hong Kong Shenzhen China

To support multiple on-demand services over fixed communication networks, network operators must allow flexible customization and fast provision of their network resources. One effective approach to this end is network virtualization, whereby each service is mapped to a virtual subnetwork providing dedicated on-demand support to network users. In practice, each service consists of a prespecified sequence of functions, called a service function chain (SFC), while each service function in a SFC can only be provided by some given network nodes. Thus, to support a given service, we must select network function nodes according to the SFC and determine the routing strategy through the function nodes in a specified order. A crucial network slicing problem that needs to be addressed is how to optimally localize the service functions in a physical network as specified by the SFCs, subject to link and node capacity constraints. In this paper, we formulate the network slicing problem as a mixed binary linear program and establish its strong NP-hardness. Furthermore, we propose efficient penalty successive upper bound minimization (PSUM) and PSUM-R(ounding) algorithms, and two heuristic algorithms to solve the problem. Simulation results are shown to demonstrate the effectiveness of the proposed algorithms. Copyright © 2017, The Authors. All rights reserved.

关键词： Network function virtualization

Semi-autonomous greenhouse environment control by combining expert knowledge and machine learning

学校读者我要写书评

暂无评论

Semi-autonomous greenhouse environment control by combining ...

Chinese Automation Congress (CAC)

作者： Mengzhen Kang Yuchen Weng Haitong Pang Liangde Li Xing-Rong Fan Jing Hua Fangle Chang Xiujuan Wang Longhua Ma The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China Ningbo Institute of Technology Zhejiang University Ningbo China The School of Computer Science and Information Engineering Chongqing Technology and Business University Chongqing China The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China Qingdao AgriTech Co. Ltd Qingdao Qingdao China The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing Engineering Research Center of Intelligent Systems and Technology Beijing China

ISBN: (数字)9781728176871

ISBN: (纸本)9781728176888

The concept of "autonomous greenhouse" hints the automatic setting of control target, and giving orders on facilities to reach these targets (set point). Information and communication technologies are widely used, but how to integrate common sense or knowledge in decision-making is still challenging. Semi-autonomous greenhouse control is to set rules with joint advantages of experienced grower and powerful machinery, with the support of knowledge graph and semantic analysis. The contribution of each side is dependent on the intelligence level of computers, the availability of data, etc. In this work, the hybrid system between human and machinery is shown with the temperature control in greenhouse. The target is set mainly by expert knowledge, while the temperature controlled by vent opening is decided by an algorithm. Result shows that the target is better reached compared to that of the experience-based approach. This example shows the feasibility of human-machine hybrid intelligence system for greenhouse environment control. Although this approach is tested for temperature control, it can be extended to control multiple factors simultaneously (temperature, light intensity, humidity, etc.) by driving multiple facilities (heating, lightening, venting, etc.).

关键词： Green products Temperature measurement Monitoring Temperature sensors Vents Agriculture Temperature distribution