Discovering new robust local search algorithms with neuro-evolution

2025-01-08

AI Summary

Existing local search strategies for black-box optimization rely heavily on handcrafted heuristics, exhibiting poor generalizability and robustness across diverse problem landscapes. Method: We propose a neural-evolutionary framework for automated algorithm discovery. It is the first to embed neural networks directly into the core neighborhood selection loop of local search, enabling end-to-end learning of transition policies. To enhance robustness, we design objective-function features invariant under monotonic transformations. Benchmarking leverages tunable NK landscapes—parameterized by ruggedness and scale—to systematically assess algorithmic performance. Contribution/Results: The discovered algorithms significantly outperform classical local search (e.g., hill climbing) and state-of-the-art learned baselines across ruggedness-varying NK landscapes, achieving faster convergence and higher solution quality. This work establishes a novel paradigm for automating the design of general-purpose, robust metaheuristic algorithms.

Technology Category

• Search and Optimization: Local Search
• Machine Learning: Evolutionary Learning
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper explores a novel approach aimed at overcoming existing challenges in the realm of local search algorithms. Our aim is to improve the decision process that takes place within a local search algorithm so as to make the best possible transitions in the neighborhood at each iteration. To improve this process, we propose to use a neural network that has the same input information as conventional local search algorithms. In this paper, which is an extension of the work [Goudet et al. 2024] presented at EvoCOP2024, we investigate different ways of representing this information so as to make the algorithm as efficient as possible but also robust to monotonic transformations of the problem objective function. To assess the efficiency of this approach, we develop an experimental setup centered around NK landscape problems, offering the flexibility to adjust problem size and ruggedness. This approach offers a promising avenue for the emergence of new local search algorithms and the improvement of their problem-solving capabilities for black-box problems.

Problem

• Local Search Algorithms
• Optimal Decision Making
• NK Landscapes

Innovation

• Neural Networks
• Local Search Algorithm Optimization
• Adaptability Enhancement