Transferring Deep Reinforcement Learning with Adversarial Objective and Augmentation

TL;DR

This paper introduces a method for deep reinforcement learning that leverages adversarial objectives and data augmentation to improve transferability and generalization across tasks, reducing catastrophic forgetting.

Contribution

It proposes a novel transfer learning approach combining adversarial training and augmentation to enhance generalization in deep reinforcement learning.

Findings

Outperforms baseline methods on Atari benchmarks

Reduces catastrophic forgetting in sequential tasks

Enhances learning speed with semi-supervised techniques

Abstract

In the past few years, deep reinforcement learning has been proven to solve problems which have complex states like video games or board games. The next step of intelligent agents would be able to generalize between tasks, and using prior experience to pick up new skills more quickly. However, most reinforcement learning algorithms for now are often suffering from catastrophic forgetting even when facing a very similar target task. Our approach enables the agents to generalize knowledge from a single source task, and boost the learning progress with a semisupervised learning method when facing a new task. We evaluate this approach on Atari games, which is a popular reinforcement learning benchmark, and show that it outperforms common baselines based on pre-training and fine-tuning.