A Dirichlet Process Mixture of Robust Task Models for Scalable Lifelong Reinforcement Learning

TL;DR

This paper introduces a scalable lifelong reinforcement learning approach using a Dirichlet process mixture model that dynamically expands and adapts to new tasks, preventing forgetting and improving generalization.

Contribution

The paper proposes a non-parametric Bayesian framework for lifelong RL that automatically adjusts model complexity and clusters tasks without explicit boundaries or heuristics.

Findings

Outperforms existing methods in robot navigation and locomotion tasks.

Effectively prevents catastrophic forgetting in lifelong learning.

Demonstrates scalable adaptation to non-stationary task distributions.

Abstract

While reinforcement learning (RL) algorithms are achieving state-of-the-art performance in various challenging tasks, they can easily encounter catastrophic forgetting or interference when faced with lifelong streaming information. In the paper, we propose a scalable lifelong RL method that dynamically expands the network capacity to accommodate new knowledge while preventing past memories from being perturbed. We use a Dirichlet process mixture to model the non-stationary task distribution, which captures task relatedness by estimating the likelihood of task-to-cluster assignments and clusters the task models in a latent space. We formulate the prior distribution of the mixture as a Chinese restaurant process (CRP) that instantiates new mixture components as needed. The update and expansion of the mixture are governed by the Bayesian non-parametric framework with an expectation maximization (EM) procedure, which dynamically adapts the model complexity without explicit task boundaries or heuristics. Moreover, we use the domain randomization technique to train robust prior parameters for the initialization of each task model in the mixture, thus the resulting model can better generalize and adapt to unseen tasks. With extensive experiments conducted on robot navigation and locomotion domains, we show that our method successfully facilitates scalable lifelong RL and outperforms relevant existing methods.