Reinforcement learning in the presence of rare events

Authors:
Jordan Frank;Shie Mannor;Doina Precup
Affiliations:
McGill University, Montreal, Quebec, Canada;McGill University, Montreal, Quebec, Canada;McGill University, Montreal, Quebec, Canada
Venue:
Proceedings of the 25th international conference on Machine learning
Year:
2008

Citing 10
Cited 1

Reinforcement Learning

Reinforcement Learning
Neuro-Dynamic Programming

Neuro-Dynamic Programming
Learning to Predict by the Methods of Temporal Differences

Machine Learning
Off-Policy Temporal Difference Learning with Function Approximation

ICML '01 Proceedings of the Eighteenth International Conference on Machine Learning
Eligibility Traces for Off-Policy Policy Evaluation

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
The Cross Entropy Method: A Unified Approach To Combinatorial Optimization, Monte-carlo Simulation (Information Science and Statistics)

The Cross Entropy Method: A Unified Approach To Combinatorial Optimization, Monte-carlo Simulation (Information Science and Statistics)
Bias and Variance Approximation in Value Function Estimates

Management Science
Adaptive Importance Sampling Technique for Markov Chains Using Stochastic Approximation

Operations Research
STEWARD: demo of spatio-textual extraction on the web aiding the retrieval of documents

dg.o '07 Proceedings of the 8th annual international conference on Digital government research: bridging disciplines & domains
Infinite-horizon policy-gradient estimation

Journal of Artificial Intelligence Research

Real-time reinforcement learning by sequential Actor-Critics and experience replay

Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

We consider the task of reinforcement learning in an environment in which rare significant events occur independently of the actions selected by the controlling agent. If these events are sampled according to their natural probability of occurring, convergence of conventional reinforcement learning algorithms is likely to be slow, and the learning algorithms may exhibit high variance. In this work, we assume that we have access to a simulator, in which the rare event probabilities can be artificially altered. Then, importance sampling can be used to learn with this simulation data. We introduce algorithms for policy evaluation, using both tabular and function approximation representations of the value function. We prove that in both cases, the reinforcement learning algorithms converge. In the tabular case, we also analyze the bias and variance of our approach compared to TD-learning. We evaluate empirically the performance of the algorithm on random Markov Decision Processes, as well as on a large network planning task.