Active Imitation Learning with Noisy Guidance

被引:0
|
作者
Brantley, Kiante [1 ]
Sharaf, Amr [1 ]
Daume, Hal, III [1 ,2 ]
机构
[1] Univ Maryland, College Pk, MD 20742 USA
[2] Microsoft Res, Redmond, WA USA
基金
美国国家科学基金会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Imitation learning algorithms provide state-of-the-art results on many structured prediction tasks by learning near-optimal search policies. Such algorithms assume training-time access to an expert that can provide the optimal action at any queried state; unfortunately, the number of such queries is often prohibitive, frequently rendering these approaches impractical. To combat this query complexity, we consider an active learning setting in which the learning algorithm has additional access to a much cheaper noisy heuristic that provides noisy guidance. Our algorithm, LEAQI, learns a difference classifier that predicts when the expert is likely to disagree with the heuristic, and queries the expert only when necessary. We apply LEAQI to three sequence labeling tasks, demonstrating significantly fewer queries to the expert and comparable (or better) accuracies over a passive approach.
引用
收藏
页码:2093 / 2105
页数:13
相关论文
共 50 条
  • [1] Robust Imitation Learning from Noisy Demonstrations
    Tangkaratt, Voot
    Charoenphakdee, Nontawat
    Sugiyama, Masashi
    [J]. 24TH INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS (AISTATS), 2021, 130 : 298 - +
  • [2] Programmatic Imitation Learning From Unlabeled and Noisy Demonstrations
    Xin, Jimmy
    Zheng, Linus
    Rahmani, Kia
    Wei, Jiayi
    Holtz, Jarrett
    Dillig, Isil
    Biswas, Joydeep
    [J]. IEEE ROBOTICS AND AUTOMATION LETTERS, 2024, 9 (06): : 4894 - 4901
  • [3] Noisy Bayesian Active Learning
    Naghshvar, Mohammad
    Javidi, Tara
    Chaudhuri, Kamalika
    [J]. 2012 50TH ANNUAL ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING (ALLERTON), 2012, : 1626 - 1633
  • [4] Active Imitation Learning of Hierarchical Policies
    Hamidi, Mandana
    Tadepalli, Prasad
    Goetschalckx, Robby
    Fern, Alan
    [J]. PROCEEDINGS OF THE TWENTY-FOURTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE (IJCAI), 2015, : 3554 - 3560
  • [5] Product image recognition with guidance learning and noisy supervision
    Li, Qing
    Peng, Xiaojiang
    Cao, Liangliang
    Du, Wenbin
    Xing, Hao
    Qiao, Yu
    Peng, Qiang
    [J]. COMPUTER VISION AND IMAGE UNDERSTANDING, 2020, 196
  • [6] Active Learning for Graphs with Noisy Structures
    Chi, Hongliang
    Qi, Cong
    Wang, Suhang
    Ma, Yao
    [J]. PROCEEDINGS OF THE 2024 SIAM INTERNATIONAL CONFERENCE ON DATA MINING, SDM, 2024, : 262 - 270
  • [7] Active Imitation Learning: Formal and Practical Reductions to IID Learning
    Judah, Kshitij
    Fern, Alan P.
    Dietterich, Thomas G.
    Tadepalli, Prasad
    [J]. JOURNAL OF MACHINE LEARNING RESEARCH, 2014, 15 : 3925 - 3963
  • [8] Active Learning from Noisy and Abstention Feedback
    Yan, Songbai
    Chaudhuri, Kamalika
    Javidi, Tara
    [J]. 2015 53RD ANNUAL ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING (ALLERTON), 2015, : 1352 - 1357
  • [9] Active Learning with Imbalanced Multiple Noisy Labeling
    Zhang, Jing
    Wu, Xindong
    Sheng, Victor S.
    [J]. IEEE TRANSACTIONS ON CYBERNETICS, 2015, 45 (05) : 1081 - 1093
  • [10] Active Inference Integrated With Imitation Learning for Autonomous Driving
    Nozari, Sheida
    Krayani, Ali
    Marin-Plaza, Pablo
    Marcenaro, Lucio
    Gomez, David Martin
    Regazzoni, Carlo
    [J]. IEEE ACCESS, 2022, 10 : 49738 - 49756