Learning with Value-Ramp

TL;DR

This paper introduces the Value-Ramp learning principle where agents follow increasing value sequences to reach reward peaks, offering a simple, natural, and robust approach for reinforcement learning.

Contribution

It proposes the novel Value-Ramp algorithm based on ramp-following intuition, emphasizing simplicity and robustness with natural number implementation.

Findings

The algorithm effectively guides agents towards reward peaks.

It is easy to configure and implement.

The approach is robust across different scenarios.

Abstract

We study a learning principle based on the intuition of forming ramps. The agent tries to follow an increasing sequence of values until the agent meets a peak of reward. The resulting Value-Ramp algorithm is natural, easy to configure, and has a robust implementation with natural numbers.