Task Transfer by Preference-Based Cost Learning

TL;DR

This paper introduces a novel reinforcement learning task transfer method that uses expert preferences instead of explicit demonstrations or cost functions, enabling more practical transfer learning.

Contribution

It proposes a framework that relaxes the need for explicit demonstrations or cost functions by leveraging expert preferences and an iterative learning process.

Findings

Effective in transferring policies without explicit cost functions

Converges reliably under theoretical analysis

Shows superior performance on benchmark tasks

Abstract

The goal of task transfer in reinforcement learning is migrating the action policy of an agent to the target task from the source task. Given their successes on robotic action planning, current methods mostly rely on two requirements: exactly-relevant expert demonstrations or the explicitly-coded cost function on target task, both of which, however, are inconvenient to obtain in practice. In this paper, we relax these two strong conditions by developing a novel task transfer framework where the expert preference is applied as a guidance. In particular, we alternate the following two steps: Firstly, letting experts apply pre-defined preference rules to select related expert demonstrates for the target task. Secondly, based on the selection result, we learn the target cost function and trajectory distribution simultaneously via enhanced Adversarial MaxEnt IRL and generate more trajectories by the learned target distribution for the next preference selection. The theoretical analysis on the distribution learning and convergence of the proposed algorithm are provided. Extensive simulations on several benchmarks have been conducted for further verifying the effectiveness of the proposed method.