Meta-Reinforcement Learning with Discrete World Models for Adaptive Load Balancing

2025/03/11 by Cameron Redovian, Redovian, Cameron
Engineering · #Artificial Intelligence (cs.AI) #D.4.1 #Elevator Systems and Control #FOS: Computer and information sciences #I.2.6 #I.2.8 #Machine Learning (cs.LG) #Operating Systems (cs.OS) #Scheduling and Optimization Algorithms

paper · pdf · doi:10.48550/arxiv.2503.08872

openalex publication_date 2025/03/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We integrate a meta-reinforcement learning algorithm with the DreamerV3 architecture to improve load balancing in operating systems. This approach enables rapid adaptation to dynamic workloads with minimal retraining, outperforming the Advantage Actor-Critic (A2C) algorithm in standard and adaptive trials. It demonstrates robust resilience to catastrophic forgetting, maintaining high performance under varying workload distributions and sizes. These findings have important implications for optimizing resource management and performance in modern operating systems. By addressing the challenges posed by dynamic and heterogeneous workloads, our approach advances the adaptability and efficiency of reinforcement learning in real-world system management tasks.

Related

• Performance-Weighed Policy Sampling for Meta-Reinforcement Learning
• Reinforcement Learning-Based Adaptive Load Balancing for Dynamic Cloud Environments
• AMAGO-2: Breaking the Multi-Task Barrier in Meta-Reinforcement Learning with Transformers
• Analysis of Reinforcement Learning for determining task replication in workflows
• A Tutorial on Meta-Reinforcement Learning