中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 自动化研究所
  3. 中国科学院自动化研究所
  4. 多模态人工智能系统全国重点实验室
  5. 机器人应用与理论组
State Primitive Learning to Overcome Catastrophic Forgetting in Robotics

文献类型：期刊论文

作者Xiong, Fangzhou1,2; Liu, Zhiyong1,2,4; Huang, Kaizhu3; Yang, Xu1; Qiao, Hong1,2,4
刊名COGNITIVE COMPUTATION
出版日期2020-11-09
页码9
关键词Catastrophic forgetting State primitives Robotics Continual learning
ISSN号1866-9956
DOI10.1007/s12559-020-09784-8
通讯作者Liu, Zhiyong(zhiyong.liu@ia.ac.cn)
英文摘要People can learn continuously a wide range of tasks without catastrophic forgetting. To mimic this functioning of continual learning, current methods mainly focus on studying a one-step supervised learning problem, e.g., image classification. They aim to retain the performance of previous image classification results when neural networks are sequentially trained on new images. In this paper, we concentrate on solving multi-step robotic tasks sequentially with the proposed architecture called state primitive learning. By projecting the original state space into a low-dimensional representation, meaningful state primitives can be generated to describe tasks. Under two kinds of different constraints on the generation of state primitives, control signals corresponding to different robotic tasks can be separately addressed only with an efficient linear regression. Experiments on several robotic manipulation tasks demonstrate the new method efficacy to learn control signals under the scenario of continual learning, delivering substantially improved performance over the other comparison methods.
资助项目National Key Research and Development Plan of China[2017YFB1300202] ; NSFC[U1613213] ; NSFC[61375005] ; NSFC[61503383] ; NSFC[61210009] ; NSFC[61876155] ; Strategic Priority Research Program of Chinese Academy of Science[XDB32050100] ; Dongguan core technology research frontier project[2019622101001] ; Natural Science Foundation of Jiangsu Province[BK20181189] ; Strategic Priority Research Program of the CAS[XDB02080003] ; Key Program Special Fund in XJTLU[KSF-A-01] ; Key Program Special Fund in XJTLU[KSF-E-26] ; Key Program Special Fund in XJTLU[KSF-P-02]
WOS研究方向Computer Science ; Neurosciences & Neurology
语种英语
WOS记录号WOS:000587981300002
出版者SPRINGER
资助机构National Key Research and Development Plan of China ; NSFC ; Strategic Priority Research Program of Chinese Academy of Science ; Dongguan core technology research frontier project ; Natural Science Foundation of Jiangsu Province ; Strategic Priority Research Program of the CAS ; Key Program Special Fund in XJTLU
源URL[http://ir.ia.ac.cn/handle/173211/41746]  
专题自动化研究所_复杂系统管理与控制国家重点实验室_机器人应用与理论组
通讯作者Liu, Zhiyong
作者单位1.Chinese Acad Sci, State Key Lab Management & Control Complex Syst, Inst Automat, Beijing 100190, Peoples R China
2.Univ Chinese Acad Sci UCAS, Sch Artificial Intelligence, Beijing 100049, Peoples R China
3.Xian Jiaotong Liverpool Univ, Dept EEE, Renai Rd 111, Suzhou 215123, Jiangsu, Peoples R China
4.CAS Ctr Excellence Brain Sci & Intelligence Techn, Shanghai 200031, Peoples R China
推荐引用方式
GB/T 7714		 Xiong, Fangzhou,Liu, Zhiyong,Huang, Kaizhu,et al. State Primitive Learning to Overcome Catastrophic Forgetting in Robotics[J]. COGNITIVE COMPUTATION,2020:9.
APA Xiong, Fangzhou,Liu, Zhiyong,Huang, Kaizhu,Yang, Xu,&Qiao, Hong.(2020).State Primitive Learning to Overcome Catastrophic Forgetting in Robotics.COGNITIVE COMPUTATION,9.
MLA Xiong, Fangzhou,et al."State Primitive Learning to Overcome Catastrophic Forgetting in Robotics".COGNITIVE COMPUTATION (2020):9.

入库方式： OAI收割

来源：自动化研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。