Strategy Extraction for Transfer in AI Agents

We propose an approach to knowledge transfer for improved lifelong learning in AI agents, using behavioural strategies as a form of transferable knowledge, influenced by the human cognitive ability to develop strategies. A strategy is defined as a partial sequence of actions an agent can take to reach some predefined event of interest. This information acts as guidance or a partial solution that an agent can generalise and use to predict how to handle unknown observed phenomena. As a first step toward this goal, we present an approach for extracting strategies from an agent's existing knowledge that can be applied in multiple contexts. Our approach uses a combination of observed action frequency information with local sequence alignment techniques to find patterns of significance that form a strategy. We demonstrate our approach in two environments: Pacman; and a dungeon-crawling video game. Our evaluation serves as a promising first step towards efficient and robust generalisation to support lifelong learning across a wider class of tasks.