Reinforcement Learning Agents with Generalizing Behavior

• DOI: 10.32473/flairs.37.1.135591
• Publication Date: 2024-07-05T19:32:19Z
• Authors: Sarah Kitchen; Reid Sawtell; Anthony Chavez; Timothy Aris
• Keywords: Generality; Traverse; Position (finance)

We explore the generality of Reinforcement Learning (RL) agents on unseen environment configurations by analyzing behavior an agent tasked with traversing a graph based from starting position to goal position. find that training single task is likely result in inflexible policies do not respond well change. Instead, wide variety scenarios offers best chance developing flexible policy, at expense increased difficulty.