Exploration in Metric State Spaces

Citation:

S. Kakade, M. Kearns, and J. Langford, Exploration in Metric State Spaces. Proceedings of the 20th International Conference on Machine Learning: , 2003.

Abstract:

We present metric- E3a provably near-optimal algorithm for reinforcement learning in Markov decisionprocesses in which there is a naturalmetricon the state space that allows the construction of accurate localmodels. The algorithm is a generalization of the E3algorithm of Kearns and Singh, and assumes a black boxfor approximate planning. Unlike the original E3, metric-E3finds a near optimal policy in an amount of timethat does not directly depend on the size of the state space, but instead depends on the covering number of thestate space. Informally, the covering number is the number of neighborhoods required for accurate localmodeling.

Publisher's Version

See also: 2003
Last updated on 10/15/2021