LGAIFeb 23, 2022

Learning Relative Return Policies With Upside-Down Reinforcement Learning

arXiv:2202.12742v2
Originality Incremental advance
AI Analysis

This is an incremental advancement for reinforcement learning researchers, demonstrating the method's potential under more complex command structures.

The paper tackled the problem of using upside-down reinforcement learning to learn policies that follow commands specifying a desired relationship between a scalar value and observed return, showing it works online in a tabular bandit setting and in CartPole with non-linear function approximation.

Lately, there has been a resurgence of interest in using supervised learning to solve reinforcement learning problems. Recent work in this area has largely focused on learning command-conditioned policies. We investigate the potential of one such method -- upside-down reinforcement learning -- to work with commands that specify a desired relationship between some scalar value and the observed return. We show that upside-down reinforcement learning can learn to carry out such commands online in a tabular bandit setting and in CartPole with non-linear function approximation. By doing so, we demonstrate the power of this family of methods and open the way for their practical use under more complicated command structures.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes