Noisy Agents: Self-supervised Exploration by Predicting Auditory Events
This work addresses the challenge of efficient exploration in reinforcement learning for agents, offering a domain-specific improvement that is incremental but effective.
The paper tackles the problem of improving exploration in reinforcement learning by proposing a novel intrinsic motivation based on predicting auditory events, resulting in significant performance gains over state-of-the-art baselines on Atari games and the emergence of physical interaction behaviors in a physics environment.
Humans integrate multiple sensory modalities (e.g. visual and audio) to build a causal understanding of the physical world. In this work, we propose a novel type of intrinsic motivation for Reinforcement Learning (RL) that encourages the agent to understand the causal effect of its actions through auditory event prediction. First, we allow the agent to collect a small amount of acoustic data and use K-means to discover underlying auditory event clusters. We then train a neural network to predict the auditory events and use the prediction errors as intrinsic rewards to guide RL exploration. Experimental results on Atari games show that our new intrinsic motivation significantly outperforms several state-of-the-art baselines. We further visualize our noisy agents' behavior in a physics environment and demonstrate that our newly designed intrinsic reward leads to the emergence of physical interaction behaviors (e.g. contact with objects).