Interpretable Reinforcement Learning for Load Balancing using Kolmogorov-Arnold Networks

TL;DR

This paper introduces an interpretable reinforcement learning approach for load balancing in networks using Kolmogorov-Arnold Networks, enabling extraction of controller equations and better understanding of decision processes.

Contribution

The paper proposes using Kolmogorov-Arnold Networks within RL to achieve interpretability in network load balancing policies, which is a novel approach.

Findings

Effective load balancing policies learned with interpretability.

Improved network throughput and reduced delay.

Ability to extract explicit controller equations.

Abstract

Reinforcement learning (RL) has been increasingly applied to network control problems, such as load balancing. However, existing RL approaches often suffer from lack of interpretability and difficulty in extracting controller equations. In this paper, we propose the use of Kolmogorov-Arnold Networks (KAN) for interpretable RL in network control. We employ a PPO agent with a 1-layer actor KAN model and an MLP Critic network to learn load balancing policies that maximise throughput utility, minimize loss as well as delay. Our approach allows us to extract controller equations from the learned neural networks, providing insights into the decision-making process. We evaluate our approach using different reward functions demonstrating its effectiveness in improving network performance while providing interpretable policies.