EMI: Exploration with Mutual Information
This work addresses exploration challenges in reinforcement learning for continuous control and image-based tasks, but it is incremental as it builds on existing intrinsic motivation methods.
The paper tackles the problem of sparse rewards in reinforcement learning by proposing EMI, an exploration method that uses mutual information to guide exploration via predictive signals in a learned representation space, achieving competitive results on locomotion and Atari tasks.
Reinforcement learning algorithms struggle when the reward signal is very sparse. In these cases, naive random exploration methods essentially rely on a random walk to stumble onto a rewarding state. Recent works utilize intrinsic motivation to guide the exploration via generative models, predictive forward models, or discriminative modeling of novelty. We propose EMI, which is an exploration method that constructs embedding representation of states and actions that does not rely on generative decoding of the full observation but extracts predictive signals that can be used to guide exploration based on forward prediction in the representation space. Our experiments show competitive results on challenging locomotion tasks with continuous control and on image-based exploration tasks with discrete actions on Atari. The source code is available at https://github.com/snu-mllab/EMI .