Mastering Rate based Curriculum Learning

TL;DR

This paper introduces a new curriculum learning algorithm based on mastering rate, which improves upon existing learning progress methods by addressing their shortcomings and achieving higher sample efficiency.

Contribution

It proposes a simpler, improved curriculum learning algorithm and a novel mastering rate concept that outperforms traditional learning progress-based methods.

Findings

Mastering rate-based algorithm outperforms learning progress methods.

Addresses shortcomings of learning progress in curriculum learning.

Significantly improves sample efficiency.

Abstract

Recent automatic curriculum learning algorithms, and in particular Teacher-Student algorithms, rely on the notion of learning progress, making the assumption that the good next tasks are the ones on which the learner is making the fastest progress or digress. In this work, we first propose a simpler and improved version of these algorithms. We then argue that the notion of learning progress itself has several shortcomings that lead to a low sample efficiency for the learner. We finally propose a new algorithm, based on the notion of mastering rate, that significantly outperforms learning progress-based algorithms.