fantastau/hh-rl

Reinforcement Learning agents for a selection Hyper-Heuristic

Hyper-Heuristic with Reinforcement Learning

This is an ongoing project as part of my PhD research, where I am implementing Reinforcement Learning agents for selecting low-level heuristics applied on combinatorial optimization problems.

Implemented agents

• Deep Q-Network
• Dynamic Multi-Armed Bandit
• Fitness-Rate-Rank Multi Armed Bandit

Combinatorial Problem Domains

The domains come from the HyFlex Framework:

• Bin Packing
• MAX-SAT
• Personnel Scheduling
• Flow Shop
• Traveling Salesman
• Vehicle Routing