Interactive Policy Learning through Confidence-Based Autonomy

被引:129
|
作者
Chernova, Sonia [1 ]
Veloso, Manuela [1 ]
机构
[1] Carnegie Mellon Univ, Dept Comp Sci, Pittsburgh, PA 15213 USA
关键词
D O I
10.1613/jair.2584
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We present Confidence-Based Autonomy (CBA), an interactive algorithm for policy learning from demonstration. The CBA algorithm consists of two components which take advantage of the complimentary abilities of humans and computer agents. The first component, Confident Execution, enables the agent to identify states in which demonstration is required, to request a demonstration from the human teacher and to learn a policy based on the acquired data. The algorithm selects demonstrations based on a measure of action selection confidence, and our results show that using Confident Execution the agent requires fewer demonstrations to learn the policy than when demonstrations are selected by a human teacher. The second algorithmic component, Corrective Demonstration, enables the teacher to correct any mistakes made by the agent through additional demonstrations in order to improve the policy and future task performance. CBA and its individual components are compared and evaluated in a complex simulated driving domain. The complete CBA algorithm results in the best overall learning performance, successfully reproducing the behavior of the teacher while balancing the tradeoff between number of demonstrations and number of incorrect actions during learning.
引用
收藏
页码:1 / 25
页数:25
相关论文
共 50 条
  • [1] Confidence-based active learning
    Li, Mingkun
    Sethi, Ishwar K.
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2006, 28 (08) : 1251 - 1261
  • [2] Confidence-Based Learning in Investment Analysis
    Serradell-Lopez, Enric
    Lara-Navarra, Pablo
    Castillo-Merino, David
    Gonzalez-Gonzalez, Ines
    [J]. TECHNOLOGY ENHANCED LEARNING: QUALITY OF TEACHING AND EDUCATIONAL REFORM, 2010, 73 : 28 - +
  • [3] CobLE : Confidence-based Learning Ensembles
    Buthpitiya, Senaka
    Dey, Anind K.
    Griss, Martin
    [J]. 2014 INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE AND COMPUTATIONAL INTELLIGENCE (CSCI), VOL 1, 2014, : 386 - 391
  • [4] Confidence-based Assessment for Learning in ePortfolio Environment
    Lap Trung Nguyen
    [J]. PROCEEDINGS 2016 5TH IIAI INTERNATIONAL CONGRESS ON ADVANCED APPLIED INFORMATICS IIAI-AAI 2016, 2016, : 328 - 331
  • [5] Improving Reinforcement Learning with Confidence-Based Demonstrations
    Wang, Zhaodong
    Taylor, Matthew E.
    [J]. PROCEEDINGS OF THE TWENTY-SIXTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2017, : 3027 - 3033
  • [6] Confidence-Based Skill Reproduction Through Perturbation Analysis
    Hertel, Brendan
    Ahmadzadeh, S. Reza
    [J]. 2023 20th International Conference on Ubiquitous Robots, UR 2023, 2023, : 165 - 170
  • [7] Confidence-Based Skill Reproduction Through Perturbation Analysis
    Hertel, Brendan
    Ahmadzadeh, S. Reza
    [J]. 2023 20TH INTERNATIONAL CONFERENCE ON UBIQUITOUS ROBOTS, UR, 2023, : 165 - 170
  • [8] Safe Reinforcement Learning via Confidence-Based Filters
    Curi, Sebastian
    Lederer, Armin
    Hirche, Sandra
    Krause, Andreas
    [J]. 2022 IEEE 61ST CONFERENCE ON DECISION AND CONTROL (CDC), 2022, : 3409 - 3415
  • [9] Confidence-based Reliable Learning under Dual Noises
    Cui, Peng
    Yue, Yang
    Deng, Zhijie
    Zhu, Jun
    [J]. ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35 (NEURIPS 2022), 2022,
  • [10] Dynamic confidence-based constraint adjustment in distributional constrained policy optimization: enhancing supply chain management through adaptive reinforcement learning
    Boutyour, Youness
    Idrissi, Abdellah
    [J]. JOURNAL OF INTELLIGENT MANUFACTURING, 2024,