Case 06 · control
The Incremental Tour Planner
POINTER NETWORKS · REINFORCE · TSP
research
The travelling salesman, learned rather than solved. A Pointer-Network encoder-decoder produces a policy for tour construction.
Trained with REINFORCE to optimize the route incrementally — building an efficient tour on the fly and adapting to changes as they arrive.
apparatus
- Pointer Networks
- REINFORCE
- PyTorch