Visualizing Critic Match Loss Landscapes for Interpretation of Online Reinforcement Learning Control Algorithms

TL;DR

This paper introduces a visualization method for critic match loss landscapes in online reinforcement learning, aiding interpretation of critic behavior and stability across training stages.

Contribution

It proposes a systematic visualization and quantitative analysis framework for critic neural networks in online RL, enhancing understanding of learning dynamics.

Findings

Distinct landscape features correlate with stable and unstable learning.

Quantitative indices enable structured comparison of training outcomes.

The method is demonstrated on cart-pole and spacecraft control tasks.

Abstract

Reinforcement learning has proven its power on various occasions. However, its performance is not always guaranteed when system dynamics change. Instead, it largely relies on users' empirical experience. For reinforcement learning algorithms with an actor-critic structure, the critic neural network reflects the approximation and optimization process in the RL algorithm. Analyzing the performance of the critic neural network helps to understand the mechanism of the algorithm. To support systematic interpretation of such algorithms in dynamic control problems, this work proposes a critic match loss landscape visualization method for online reinforcement learning. The method constructs a loss landscape by projecting recorded critic parameter trajectories onto a low-dimensional linear subspace. The critic match loss is evaluated over the projected parameter grid using fixed reference state samples and temporal-difference targets. This yields a three-dimensional loss surface together with a two-dimensional optimization path that characterizes critic learning behavior. To extend analysis beyond visual inspection, quantitative landscape indices and a normalized system performance index are introduced, enabling structured comparison across different training outcomes. The approach is demonstrated using the Action-Dependent Heuristic Dynamic Programming algorithm on cart-pole and spacecraft attitude control tasks. Comparative analyses across projection methods and training stages reveal distinct landscape characteristics associated with stable convergence and unstable learning. The proposed framework enables both qualitative and quantitative interpretation of critic optimization behavior in online reinforcement learning.