Tabu Tenure Policy with Deep Reinforcement Learning
Date:
Presented research on integrating deep reinforcement learning with tabu search optimization algorithms at the International Conference on Optimization and Learning.
Presentation Overview
This presentation explores novel methods for dynamically adjusting tabu search parameters using deep reinforcement learning. The approach learns optimal tenure policies during algorithm execution rather than using fixed parameter configurations.
Key Points
- Integration of deep RL with classical tabu search algorithms
- Dynamic parameter adjustment during optimization
- Significant performance improvements over static configurations
- Real-world applications in discrete optimization
Contribution
The work bridges classical metaheuristics with modern deep learning approaches, demonstrating that learned parameter policies can outperform manually tuned configurations.