检索结果-内蒙古大学图书馆

25th International Conference on Neural Information Processing, ICONIP 2018

作者： Xiong, Fangzhou Liu, Zhiyong Yang, Xu The State Key Lab of Management and Control for Complex Systems Institute of Automation Chinese Academy of Science Beijing100190 China Beijing100049 China Centre for Excellence in Brain Science and Intelligence Technology Chinese Academy of Sciences Shanghai200031 China Cloud Computing Center Chinese Academy of Sciences DongGuanGuangDong523808 China

ISBN: (纸本)9783030041816

Catastrophic forgetting is a tough issue when the agent faces the sequential multi-task learning scenario without storing previous task information. It gradually becomes an obstacle to achieve artificial general intelligence which is generally believed to behave like a human with continuous learning capability. In this paper, we propose to utilize the variational Bayesian inference method to overcome catastrophic forgetting. By pruning the neural network according to the mean and variance of weights, parameters are vastly reduced, which mitigates the storage problem of double parameters required in variational Bayesian inference. Based on this lightweight version, autoencoders trained on different tasks are employed to self-adaptively match the corresponding task parameters to tackle sequential multi-task learning problem. We show experimentally on several fundamental datasets that the proposed method can perform substantial improvements without catastrophic forgetting over other classic methods especially in the setting where the probability distributions between tasks present more different. © Springer Nature Switzerland AG 2018.

关键词： Probability distributions

来源：评论

学校读者我要写书评

暂无评论

Research on Social Marketing Strategies with An Agent-based Propagation Model

引用

IFAC-PapersOnLine 2017年第1期50卷 13581-13586页

作者： Zeng, Shuai Li, Juanjuan Ni, Xiaochun Yuan, Yong Wang, Fei-Yue The State Key Lab of Management Control for Complex Systems Institute of Automation Chinese Academy of Sciences China Qingdao Academy of Intelligent Industries BeijingShandong China Research Center of Military Computational Experiments and Parallel Systems National University of Defense Technology Changsha China

Considering the increasing complexity of social networks and user behaviors, it is very challenging for advertisers to formulate their strategies of selecting proper initial seed users in their social marketing efforts. In this paper, we tackle this challenge by proposing an agent-based propagation model and injecting it into typical social networks with three types of structures, i.e., Erds-Rényi random graph, Watt-Strogatz small world graph, and Barabási-Albert scale-free graph. We instantiate this agent-based model with demographic characteristics extracted from real-world census data collected in China. By investigating the diffusion process of advertising information in these social networks, we can analyze and compare the performance of advertisers’ targeting and influencer strategies. Our experimental results indicate that advertisers adopting influencer strategies should manipulate the initial well-connected seeds to deliver information only to the potential customers instead of a wide range of generic users. © 2017

关键词： Marketing Autonomous agents Behavioral research Commerce Computational methods Population statistics Agent based model Demographic characteristics Information diffusion Potential customers Propagation modeling Seeding strategies Small world graphs Social marketings

来源：评论

学校读者我要写书评

暂无评论

A Projection-Based Algorithm for Constrained L1- Minimization Optimization with Application to Sparse Signal Reconstruction

A Projection-Based Algorithm for Constrained L1- Minimizatio...

引用

International Conference on Information Science and Technology (ICIST)

作者： Qingshan Liu Wei Zhang Jiang Xiong Bingrong Xu Long Cheng School of Mathematics Southeast University Nanjing China College of Computer Science and Engineering Chongqing Three Gorges University Chongqing China School of Automation Huazhong University of Science and Technolog< Wuhan China State Key Lab. of Management and Control for Complex Systems University of Chinese Academy of Sciences Beijing China

In this paper, a projection-base algorithm is proposed for solving the constrained L 1 -minimization problem. Furthermore, the algorithm is utilized to sparse signal reconstruction. The L 1 -minimization is first converted into some equations which are described by the projections onto a hyper box set and the nonnegative quadrant. Then a iterative algorithm is proposed for solving the L 1 -minimization problem. Next, the algorithm is applied to sparse signal reconstruction described as an L 1 -minimization problem subject to L ∞ -norm noise constraint, or equivalently bound constraint. Finally, several experiments are presented to show the performance of the proposed algorithm.

关键词： Signal processing algorithms Signal reconstruction Optimization Convergence Electronic mail Recurrent neural networks Lyapunov methods

来源：评论

学校读者我要写书评

暂无评论

An Operation management Cloud Ecosystem for Smart Buildings Based on Internet of Things

An Operation Management Cloud Ecosystem for Smart Buildings ...

引用

IEEE International Conference on Cyber Technology in Automation, control, and Intelligent systems

作者： Ming Wang Shi Qiu Guiqing Zhang Junzhi Yu Yuling Wang School of Information and Electrical Engineering Shandong Jianzhu University State Key Lab. of Management and Control for Complex Systems Institution of Automation Chinese Academy of Sciences

ISBN: (纸本)9781538604915

It is a key issue in building operation management how to design an open and ubiquitous information system for diversity buildings of smart cities. This paper presents an operation management cloud ecosystem for smart buildings based on the technologies of Internet of Things (IoT) and cloud computing. It contains four levels: operation management application level, Internet of things platform, networking level, node level. The IoT platform provides hosting and computing, cloud storage, data storage for smart buildings. Networking level takes the responsibility for networking for intelligent hardware and smart terminals from the device level. With the building operation management applications, the cloud ecosystem is able to provide services of monitoring, controlling, management, and real-time optimization for smart buildings in cities.

关键词： intelligent buildings Internet of things organisation and administration Computer based information systems Thing Internets Clouds Building construction Networking Cloud Computing

来源：评论

学校读者我要写书评

暂无评论

Design and dynamic analysis for Amoeba-like roboţs turning-mechanism applied with spring and damp system

Design and dynamic analysis for Amoeba-like roboţs turning-...

引用

2016 International Conference on Advanced Robotics and Mechatronics, ICARM 2016

作者： Dong, Xiang Sun, Henghui Wang, Zengfu Cheng, Long Institute of Intelligent Machines Chinese Academy of Sciences Hefei230031 China State Key Lab. of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China

ISBN: (纸本)9781509033645

A new turning-mechanism for Amoeba-like Robot was proposed in this paper, which based on the amoeba-like robot kinematics characteristics of tail contracting and skin flipping. First, one kind of variable-speed node composed of the spring and damp was designed. Because of the existing of the variable-speed nodes, the velocities of the two strip-skins depended not only on the velocities of threads that gone through the variable-speed nodes, but also on the spring and damp it contained. So, although the velocities of threads that induced the velocities of the strip-skins were the same, the velocities of the strip-skins still can be different since the variable-speed nodes existed. That difference in velocities made the turning movement possible. Second, the dynamic model for the turning movement of the Amoeba-like Robot was established and analyzed. The analysis results included the angular velocity and angle curve related to time, which were computed with the dynamic model when shortening speed of threads length was constant. The results achieved in this paper provided an important basic for the research on the behaviors and control strategy of amoeba-like robot. The new turning-mechanism can be popularized to the tracked mobile robot, for it provided a new and availab.e turning method. © 2016 IEEE.

关键词： Velocity

来源：评论

学校读者我要写书评

暂无评论

Power system state estimation via feasible point pursuit: Algorithms and Cramér-Rao bound

arXiv

引用

arXiv 2017年

作者： Wang, Gang Zamzam, Ahmed S. Giannakis, Georgios B. Sidiropoulos, Nicholas D. Digital Tech. Center ECE Dept. U. of Minnesota MplsMN55455 United States State Key Lab. of Intelligent Control and Decision of Complex Systems Beijing Inst. of Tech. Beijing100081 China

Accurately monitoring the system's operating point is central to the reliable and economic operation of an electric power grid. Power system state estimation (PSSE) aims to obtain complete voltage magnitude and angle information at each bus given a number of system variables at selected buses and lines. Power flow analysis is a special case of PSSE, and amounts to solving a set of noise-free power flow equations. Physical laws dictate quadratic relationships between availab.e quantities and unknown voltages, rendering general instances of power flow and PSSE nonconvex and NP-hard. Past approaches are largely based on gradient-type iterative procedures or semidefinite relaxation (SDR). Due to nonconvexity, the solution obtained via gradienttype schemes depends on initialization, while SDR methods do not perform as desired in challenging scenarios. This paper puts forth novel feasible point pursuit (FPP)-based solvers for power flow and PSSE, which iteratively seek feasible solutions for a nonconvex quadratically constrained quadratic programming (QCQP) reformulation of the weighted least-squares (WLS) problem. Relative to the prior art, the developed solvers offer superior performance at the cost of higher complexity. Furthermore, they converge to a stationary point of the WLS problem. As a baseline for comparing different estimators, the Cramér-Rao lower bound (CRLB) is derived for the fundamental PSSE problem in this paper. Judicious numerical tests on several IEEE benchmark systems showcase markedly improved performance of our FPPbased solvers for both power flow and PSSE tasks over popular WLS-based Gauss-Newton iterations and SDR approaches. Copyright © 2017, The Authors. All rights reserved.

关键词： state estimation

来源：评论

学校读者我要写书评

暂无评论

7.3: Compare Variation of User's EEG Owing to Parallax Cue

引用

SID Symposium Digest of Technical Papers 2018年第S1期49卷

作者： Haichen Hu Danli Wang Kang Yue Xinpan Yang Steve C. Chiu Beijing Key Lab of Human-Computer Interaction Institute of Software Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China The State Key Laboratory of Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China College of Science and Engineering Idaho State University Pocatello America

3D display has become increasingly popular in many fields. It can provide incredible immersion compared with 2D display. There are many researches referring to the change of user about no matter subjective experience or objective physiology signal while watching 3D content. Similarly, in this paper, we study the variation of user’s EEG when the user watched 3D content. However, we only focus on parallax cue and avoid other cues of 3D content by using Random Dot Stereogram(RDS) as visual stimulus. We designed an experiment which asked the subject to watch RDS with nine kinds of parallaxes(0,±0.5,±1.1,±1.3,±1.7). The users watched one RDS for 1 minute and then closed eyes and rest for 1 minute every time. During watching, EEG signal with 30-channel was recorded. The result suggests that significant variation of the user’s EEG data occurred in the Alpha (8–13Hz) and Beta(14–30Hz) frequency band by different parallax’s comparison. And we found not only parallax but also electrode position can impact the user’s EEG data significantly. Accurately, the EEG data has a significant change in six electrodes position in both frequency bands including CZ, PZ, OZ, O2 in international 10–20 system. It means those electrode positions we detect can be regarded as typical positions where we record EEG signal. It will simplify the process of EEG signal acquisition when we need to study the relationship between parallax cue of 3D display and the user further.

关键词： EEG Parallax RDS 3D

来源：评论

学校读者我要写书评

暂无评论

Analyse Social Influence on Student Motivation based on Social Activity Network

Analyse Social Influence on Student Motivation based on Soci...

引用

2016年IEEE/INFORMS服务运筹、物流与信息化国际会议

作者： Shuai ZENG Jie ZHANG The State Key Lab of Management and Control for Complex Systems Institute of AutomationChinese Academy of Sciences Qingdao Academy of Intelligent Industries

Student motivation is one of the most important individual characters for explaining student performance in class. Existing researches show that it can be affected largely by social-contextual factors. Rather than focusing on social relationship, in this paper we construct social activity networks for analyzing the impact of collective behaviors on student motivation. We conduct field experiments to compare social activity networks with relationship networks by analyzing the structural characteristics. We also investigate the dynamic of social activities,and validate the effectiveness of social activity network in reflecting social influence on student motivation. Our results show that social activities are important factors affecting student motivation.

关键词： student motivation collective behaviour so cial network analysis online course

来源：评论

学校读者我要写书评

暂无评论

Design and dynamic analysis for Amoeba-like robot's turning-mechanism applied with spring and damp system

Design and dynamic analysis for Amoeba-like robot's turning-...

引用

International Conference on Advanced Robotics and Mechatronics (ICARM)

作者： Xiang Dong HengHui Sun Zeng Fu Wang Long Cheng Institute of Intelligent Machines Chinese Academy of Sciences Hefei Anhui CN The State Key Lab. of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing Beijing CN

ISBN: (纸本)9781509033652

A new turning-mechanism for Amoeba-like Robot was proposed in this paper, which based on the amoeba-like robot kinematics characteristics of tail contracting and skin flipping. First, one kind of variable-speed node composed of the spring and damp was designed. Because of the existing of the variable-speed nodes, the velocities of the two strip-skins depended not only on the velocities of threads that gone through the variable-speed nodes, but also on the spring and damp it contained. So, although the velocities of threads that induced the velocities of the strip-skins were the same, the velocities of the strip-skins still can be different since the variable-speed nodes existed. That difference in velocities made the turning movement possible. Second, the dynamic model for the turning movement of the Amoeba-like Robot was established and analyzed. The analysis results included the angular velocity and angle curve related to time, which were computed with the dynamic model when shortening speed of thread's length was constant. The results achieved in this paper provided an important basic for the research on the behaviors and control strategy of amoeba-like robot. The new turning-mechanism can be popularized to the tracked mobile robot, for it provided a new and availab.e turning method.

关键词： Conferences Robots Mechatronics

来源：评论

学校读者我要写书评

暂无评论

NFLB Dropout:Improve Generalization Ability by Dropping out the Best-A Biologically Inspired Adaptive Dropout Method For Unsupervised Learning

NFLB Dropout:Improve Generalization Ability by Dropping out ...

引用

International Joint Conference on Neural Networks

作者： Peijie Yin Lu Qi Xuanyang Xi Bo Zhang Hong Qiao Institute of Applied Mathematics Academy of Mathematics and Systems Science Chinese Academy of Science State Key Lab of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences

ISBN: (纸本)9781509006212

Generalization ability is widely acknowledged as one of the most important criteria to evaluate the quality of unsupervised models. The objective of our research is to find a better dropout method to improve the generalization ability of convolutional deep belief network (CDBN), an unsupervised learning model for vision tasks. In this paper, the phenomenon of low feature diversity during the training process is investigated. The attention mechanism of human visual system is more focused on rare events and depresses well-known facts. Inspired by this mechanism, No Feature Left Behind Dropout (NFLB Dropout), an adaptive dropout method is firstly proposed to automatically adjust the dropout rate feature-wisely. In the proposed method, the algorithm drops well-trained features and keeps poorly-trained ones with a high probability during training iterations. In addition, we apply two approximations of the quality of features, which are inspired by theory of saliency and optimization. Compared with the model trained by standard dropout, experiment results show that our NFLB Dropout method improves not only the accuracy but the convergence speed as well.

关键词： Training Feature extraction Brain models Adaptation models Neural networks Unsupervised learning

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：