Material Detail

Reinforcement Learning with Limited Reinforcement: Using Bayes Risk for Active Learning in POMDPs

Reinforcement Learning with Limited Reinforcement: Using Bayes Risk for Active Learning in POMDPs

This video was recorded at 25th International Conference on Machine Learning (ICML), Helsinki 2008. Partially Observable Markov Decision Processes (POMDPs) have succeeded in planning domains because they optimally trade between actions that increase an agent's knowledge and actions that increase an agent's reward. Unfortunately, most POMDPs are defined with a large number of parameters which are difficult to specify only from domain knowledge. In this paper, we treat the POMDP model parameters as additional hidden state in a "model-uncertainty" POMDP and develop an approximate algorithm for planning in the this larger POMDP. The approximation, coupled with model-directed queries, allows the planner to actively learn good policies. We demonstrate our approach on several standard POMDP problems.

Quality

  • User Rating
  • Comments
  • Learning Exercises
  • Bookmark Collections
  • Course ePortfolios
  • Accessibility Info

More about this material

Browse...

Disciplines with similar materials as Reinforcement Learning with Limited Reinforcement: Using Bayes Risk for Active Learning in POMDPs

Comments

Log in to participate in the discussions or sign up if you are not already a MERLOT member.