Lifelong Learning using Eigentasks: Task Separation, Skill Acquisition, and Selective Transfer
This addresses the problem of catastrophic forgetting and limited knowledge transfer in lifelong learning for AI systems, though it appears incremental as it extends existing generative replay approaches.
The paper tackles lifelong learning by introducing the eigentask framework, which pairs skills with generative models to address goals like forward knowledge transfer, achieving improved performance over state-of-the-art in supervised continual learning and demonstrating forward knowledge transfer in Starcraft2 RL.
We introduce the eigentask framework for lifelong learning. An eigentask is a pairing of a skill that solves a set of related tasks, paired with a generative model that can sample from the skill's input space. The framework extends generative replay approaches, which have mainly been used to avoid catastrophic forgetting, to also address other lifelong learning goals such as forward knowledge transfer. We propose a wake-sleep cycle of alternating task learning and knowledge consolidation for learning in our framework, and instantiate it for lifelong supervised learning and lifelong RL. We achieve improved performance over the state-of-the-art in supervised continual learning, and show evidence of forward knowledge transfer in a lifelong RL application in the game Starcraft2.