Learning sparse representations in reinforcement learning with sparse coding

A variety of representation learning approaches have been investigated for reinforcement learn- ing; much less attention, however, has been given to investigating the utility of sparse coding. Outside of reinforcement learning, sparse coding representations have been widely used, with non-convex objectives that result in discriminative representations. In this work, we develop a supervised sparse coding objective for policy evaluation. Despite the non-convexity of this objective, we prove that all local minima are global minima, making the approach amenable to simple optimization strategies. We empirically show that it is key to use a supervised objective, rather than the more straightforward unsupervised sparse coding approach. We compare the learned representations to a canonical fixed sparse representation, called tile-coding, demon- strating that the sparse coding representation outperforms a wide variety of tile-coding representations.

Authors

• Raksha Kumaraswamy
• Lei Le

Keywords

No keywords are indexed for this paper.