Efficiently Computable Safety Bounds for Gaussian Processes in Active Learning

被引:0
|
作者
Tebbe, Joern [1 ]
Zimmer, Christoph [2 ]
Steland, Ansgar [3 ]
Lange-Hegermann, Markus [1 ]
机构
[1] OWL Univ Appl Sci & Arts, Lemgo, Germany
[2] Bosch Ctr Artificial Intelligence, Renningen, Germany
[3] Rhein Westfal TH Aachen, Aachen, Germany
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
Active learning of physical systems must commonly respect practical safety constraints, which restricts the exploration of the design space. Gaussian Processes (GPs) and their calibrated uncertainty estimations are widely used for this purpose. In many technical applications the design space is explored via continuous trajectories, along which the safety needs to be assessed. This is particularly challenging for strict safety requirements in GP methods, as it employs computationally expensive Monte-Carlo sampling of high quantiles. We address these challenges by providing provable safety bounds based on the adaptively sampled median of the supremum of the posterior GP. Our method significantly reduces the number of samples required for estimating high safety probabilities, resulting in faster evaluation without sacrificing accuracy and exploration speed. The effectiveness of our safe active learning approach is demonstrated through extensive simulations and validated using a real-world engine example.
引用
收藏
页数:21
相关论文
共 50 条
  • [1] Efficiently Computable Bounds for Magic State Distillation
    Wang, Xin
    Wilde, Mark M.
    Su, Yuan
    [J]. PHYSICAL REVIEW LETTERS, 2020, 124 (09)
  • [2] Computable bounds for the discrimination of Gaussian states
    Pirandola, Stefano
    Lloyd, Seth
    [J]. PHYSICAL REVIEW A, 2008, 78 (01)
  • [3] Optimism in Active Learning with Gaussian Processes
    Collet, Timothe
    Pietquin, Olivier
    [J]. NEURAL INFORMATION PROCESSING, PT II, 2015, 9490 : 152 - 160
  • [5] Upper and lower bounds on the learning curve for Gaussian processes
    Williams, CKI
    Vivarelli, F
    [J]. MACHINE LEARNING, 2000, 40 (01) : 77 - 102
  • [6] Upper and Lower Bounds on the Learning Curve for Gaussian Processes
    Christopher K.I. Williams
    Francesco Vivarelli
    [J]. Machine Learning, 2000, 40 : 77 - 102
  • [7] Safe Exploration for Active Learning with Gaussian Processes
    Schreiter, Jens
    Duy Nguyen-Tuong
    Eberts, Mona
    Bischoff, Bastian
    Markert, Heiner
    Toussaint, Marc
    [J]. MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES, PT III, 2015, 9286 : 133 - 149
  • [8] Active learning with Gaussian processes for object categorization
    Kapoor, Ashish
    Grauman, Kristen
    Urtasun, Raquel
    Darrell, Trevor
    [J]. 2007 IEEE 11TH INTERNATIONAL CONFERENCE ON COMPUTER VISION, VOLS 1-6, 2007, : 134 - +
  • [9] Promoting active learning with mixtures of Gaussian processes
    Zhao, Jing
    Sun, Shiliang
    Wang, Huijuan
    Cao, Zehui
    [J]. KNOWLEDGE-BASED SYSTEMS, 2020, 188
  • [10] Active Learning of Linear Embeddings for Gaussian Processes
    Garnett, Roman
    Osborne, Michael A.
    Hennig, Philipp
    [J]. UNCERTAINTY IN ARTIFICIAL INTELLIGENCE, 2014, : 230 - 239