Hilbert Space Embeddings of POMDPs
This work addresses policy learning for POMDPs, which is a domain-specific problem in reinforcement learning, and appears incremental as it adapts kernel methods to this context.
The authors tackled policy learning in partially observable Markov decision processes (POMDPs) by proposing a nonparametric approach using Hilbert space embeddings, and they confirmed that the correct policy is learned in experiments.
A nonparametric approach for policy learning for POMDPs is proposed. The approach represents distributions over the states, observations, and actions as embeddings in feature spaces, which are reproducing kernel Hilbert spaces. Distributions over states given the observations are obtained by applying the kernel Bayes' rule to these distribution embeddings. Policies and value functions are defined on the feature space over states, which leads to a feature space expression for the Bellman equation. Value iteration may then be used to estimate the optimal value function and associated policy. Experimental results confirm that the correct policy is learned using the feature space representation.