Curriculum Development for Transfer Learning in Dynamic Multiagent Settings
Abstract
Transfer learning in reinforcement learning has been an active area of research over the past decade. In transfer learning, training on a source task is leveraged to speed up or otherwise improve learning on a target task. This project addressed the ambitious problem of curriculum learning in reinforcement learning, in which the goal is to design a sequence of source tasks for an agent to train on, such that final performance or learning speed is improved. We take the position that each stage of such a curriculum should be tailored to the ability of the agent in order to promote learning new behaviors. To tackle the problem of curriculum learning, we addressed three key sub-problems: 1) Learning Transferability, and 2) Automatic Source Task Creation, 3) Curriculum Construction through Crowd Sourcing. This technical report documents the methods, experiments, and results of the proposed frameworks for curriculum construction for reinforcement learning agents.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2016
- Accession Number
- AD1011857
Entities
People
- Jivko Sinapov
- Matthew Taylor
- Peter Stone
Organizations
- University of Texas at Austin