Optimizing Tabu Search Parameters with Machine Learning

This paper presents novel methods for dynamically tuning tabu search algorithm parameters using machine learning techniques. Instead of using fixed parameter configurations, the approach adapts parameters based on problem-solving progress.

Key Contributions

• Adaptive parameter ML-based tuning system for tabu search
• 5% improvement in optimization performance over static parameter configurations
• Statistical significance testing of results
• Real-world validation on discrete optimization problems

Methodology

The approach uses machine learning models trained on historical optimization runs to predict optimal parameter settings during algorithm execution, enabling dynamic parameter adjustment.

Recommended citation: Konovalenko, A., & Hvattum, L.M. (2024). "Optimizing Tabu Search Parameters with Machine Learning." Logistics.
Download Paper