A Reinforcement Learning Inspired Latent Yield Based Adaptive Algorithm Switching Mechanism

2026-05-23

AI Summary

This work addresses the instability of algorithm selection in dynamic environments, where performance is highly sensitive to shifts in instance characteristics. To enhance robustness, the authors propose an adaptive switching mechanism that integrates reinforcement learning with an island-model framework. Central to this approach is a latent utility metric designed to be robust against feature perturbations, which aggregates historical performance to mitigate the impact of transient fluctuations. By combining an exploration–exploitation strategy with a parallel population-based algorithm architecture, the method enables efficient and stable algorithm switching. Empirical evaluations on sorting algorithm selection and robotic obstacle avoidance tasks demonstrate significant improvements in both performance and robustness under dynamic conditions, confirming the efficacy of the proposed approach.

Abstract

Selecting the most suitable algorithm for a given problem instance remains a challenging task, particularly in online or dynamic environments where problem characteristics evolve over time. Relying solely on instantaneous performance metrics can result in a reactive and unstable behaviour, often leading to suboptimal algorithm switching. This paper introduces a computationally efficient approach for aggregating an algorithm's performance across multiple problem instances that is fairly immune to erratic variations in instance features. Inspired by features inherent to Reinforcement Learning (RL), this technique encapsulates rewards and penalties into a latent yield that, in turn, triggers exploitation and exploration, consequently resulting in adaptive algorithm switching. The proposed technique employs island models, inspired by Genetic Algorithms, to facilitate parallel exploration and performance exchanges among algorithm populations inhabiting local repertoires. Experimental evaluations on sorting algorithms and robotic obstacle avoidance tasks demonstrate the feasibility and effectiveness of the approach, highlighting its potential in domains where adaptive algorithm selection is critical.

Problem

• algorithm selection
• dynamic environments
• adaptive switching
• performance aggregation
• online optimization

Innovation

• Reinforcement Learning
• Latent Yield
• Adaptive Algorithm Switching
• Island Models
• Exploration-Exploitation Tradeoff