Representations and Exploration for Deep Reinforcement Learning using Singular Value Decomposition
This work addresses key challenges in deep reinforcement learning for agents in complex, partially observable domains, offering a scalable solution with incremental improvements over existing methods.
The authors tackled representation learning and exploration in deep reinforcement learning by developing a singular value decomposition-based method that preserves transition structure and estimates state visitation frequencies for pseudo-counts, demonstrating effectiveness in multi-task and hard exploration tasks on DM-Lab-30 and DM-Hard-8 environments.
Representation learning and exploration are among the key challenges for any deep reinforcement learning agent. In this work, we provide a singular value decomposition based method that can be used to obtain representations that preserve the underlying transition structure in the domain. Perhaps interestingly, we show that these representations also capture the relative frequency of state visitations, thereby providing an estimate for pseudo-counts for free. To scale this decomposition method to large-scale domains, we provide an algorithm that never requires building the transition matrix, can make use of deep networks, and also permits mini-batch training. Further, we draw inspiration from predictive state representations and extend our decomposition method to partially observable environments. With experiments on multi-task settings with partially observable domains, we show that the proposed method can not only learn useful representation on DM-Lab-30 environments (that have inputs involving language instructions, pixel images, and rewards, among others) but it can also be effective at hard exploration tasks in DM-Hard-8 environments.