Guardado en:
Detalles Bibliográficos
Autores principales: Singh, Kamal, Marouani, Sami, Sheikh, Ahmad Al, Quang, Pham Tran Anh, Habrard, Amaury
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2505.14459
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910957659226112
author Singh, Kamal
Marouani, Sami
Sheikh, Ahmad Al
Quang, Pham Tran Anh
Habrard, Amaury
author_facet Singh, Kamal
Marouani, Sami
Sheikh, Ahmad Al
Quang, Pham Tran Anh
Habrard, Amaury
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2505_14459
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Interpretable Reinforcement Learning for Load Balancing using Kolmogorov-Arnold Networks
Singh, Kamal
Marouani, Sami
Sheikh, Ahmad Al
Quang, Pham Tran Anh
Habrard, Amaury
Machine Learning
Networking and Internet Architecture
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.
title Interpretable Reinforcement Learning for Load Balancing using Kolmogorov-Arnold Networks
topic Machine Learning
Networking and Internet Architecture
url https://arxiv.org/abs/2505.14459