Robust Active Learning

  • Author / Creator
    Nie, Rui
  • This dissertation first introduces the concepts of robust active learning (also called optimal experimental design in statistics), and the possible advantages of it over the traditional passive learning method. Then a general regression problem with possibly misspecified models is presented, and divided into three specific problems due to different choices of loss functions and optimizing methods. After that, the three problems are all solved with a minimax approach but in different ways to get the optimal design densities for the active learning method. Finally, simulations are used to compare active learning with passive learning results on specific examples, and the experiment results prove that active learning is more robust and advantageous than passive learning in these examples.

  • Subjects / Keywords
  • Graduation date
  • Type of Item
  • Degree
    Master of Science
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
  • Language
  • Institution
    University of Alberta
  • Degree level
  • Department
    • Department of Mathematical and Statistical Sciences
  • Specialization
    • Statistical Machine Learning
  • Supervisor / co-supervisor and their department(s)
    • Wiens, Douglas (Department of Mathematical and Statistical Sciences)
  • Examining committee members and their departments
    • Gombay, Edit (Department of Mathematical and Statistical Sciences)
    • Mizera, Ivan (Department of Mathematical and Statistical Sciences)
    • Wiens, Douglas (Department of Mathematical and Statistical Sciences)
    • Greiner, Russell (Department of Computing Science)