INVESTIGATING THE ROLE OF REINFORCEMENT LEARNING IN OPTIMIZING AUTOMATED TASK PIPELINES ACROSS HETEROGENEOUS SYSTEMS

Abstract

Reinforcement Learning (RL) has emerged as a powerful tool for optimizing decision-making processes in complex environments. This paper explores its application to automate and optimize task pipelines across heterogeneous systems, which typically face challenges due to variable performance, resource availability, and workload distribution. Through analysis of recent advancements and historical insights, we evaluate the effectiveness, efficiency, and adaptability of RL in this domain. We further propose a basic framework incorporating RL with pipeline schedulers to highlight optimization potential.