Stein Variational Goal Generation for adaptive Exploration in Multi-Goal Reinforcement Learning
arXiv:2206.06719
Abstract
In multi-goal Reinforcement Learning, an agent can share experience between related training tasks, resulting in better generalization for new tasks at test time. However, when the goal space has discontinuities and the reward is sparse, a majority of goals are difficult to reach. In this context, a curriculum over goals helps agents learn by adapting training tasks to their current capabilities. In this work we propose Stein Variational Goal Generation (SVGG), which samples goals of intermediate difficulty for the agent, by leveraging a learned predictive model of its goal reaching capabilities. The distribution of goals is modeled with particles that are attracted in areas of appropriate difficulty using Stein Variational Gradient Descent. We show that SVGG outperforms state-of-the-art multi-goal Reinforcement Learning methods in terms of success coverage in hard exploration problems, and demonstrate that it is endowed with a useful recovery property when the environment changes.
References in corpus (7)
- Reinforcement Learning with Prototypical Representations
- Maximum Entropy Gain Exploration for Long Horizon Multi-goal Reinforcement Learning
- Automatic Curriculum Learning through Value Disagreement
- Rewriting History with Inverse RL: Hindsight Inference for Policy Improvement
- Variational Automatic Curriculum Learning for Sparse-Reward Cooperative Multi-Agent Problems
- Annealed Stein Variational Gradient Descent
- Density-based Curriculum for Multi-goal Reinforcement Learning with Sparse Rewards