Contextual Bandit Learning with Predictable Rewards

Details Contextual Bandit Learning with Predictable Rewards Contextual Bandit Learning with Predictable Rewards

2012-02-07

arxiv.org/abs/1202.1334v2

Computer Science

Learning

Scientific paper

Contextual bandit learning is a reinforcement learning problem where the learner repeatedly receives a set of features (context), takes an action and receives a reward based on the action and context. We consider this problem under a realizability assumption: there exists a function in a (known) function class, always capable of predicting the expected reward, given the action and context. Under this assumption, we show three things. We present a new algorithm---Regressor Elimination--- with a regret similar to the agnostic setting (i.e. in the absence of realizability assumption). We prove a new lower bound showing no algorithm can achieve superior performance in the worst case even with the realizability assumption. However, we do show that for any set of policies (mapping contexts to actions), there is a distribution over rewards (given context) such that our new algorithm has constant regret unlike the previous approaches.

Affiliated with

Also associated with

No associations

Rating

Contextual Bandit Learning with Predictable Rewards does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Contextual Bandit Learning with Predictable Rewards, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Contextual Bandit Learning with Predictable Rewards will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-578810