Read the original document by opening this link in a new tab.
Table of Contents
1 Introduction
2 Progressive Networks
3 Transfer Analysis
4 Related Literature
5 Experiments
Summary
Learning to solve complex sequences of tasks while both leveraging transfer and avoiding catastrophic forgetting remains a key obstacle to achieving human-level intelligence. The progressive networks approach represents a step forward in this direction. Progressive networks integrate the ability to learn multiple tasks in sequence, transfer knowledge from previous tasks, and prevent catastrophic forgetting. The architecture of progressive networks includes lateral connections to access previously learned features, allowing for deep compositionality. The experimental results show that progressive networks outperform traditional baselines in terms of transfer and performance on various RL tasks.