Learning heuristics for A*

TL;DR

This paper introduces a neural network-based approach to learn heuristic functions for A* search, significantly improving pathfinding efficiency on graphs while maintaining optimality.

Contribution

It combines multi-task learning of Dijkstra's algorithm and heuristics, enabling faster A* search with learned heuristics that preserve optimal paths.

Findings

Learned heuristics speed up A* search compared to Dijkstra.

The approach maintains minimal-cost path accuracy.

Neural heuristics generalize well across graph instances.

Abstract

Path finding in graphs is one of the most studied classes of problems in computer science. In this context, search algorithms are often extended with heuristics for a more efficient search of target nodes. In this work we combine recent advancements in Neural Algorithmic Reasoning to learn efficient heuristic functions for path finding problems on graphs. At training time, we exploit multi-task learning to learn jointly the Dijkstra's algorithm and a consistent heuristic function for the A* search algorithm. At inference time, we plug our learnt heuristics into the A* algorithm. Results show that running A* over the learnt heuristics value can greatly speed up target node searching compared to Dijkstra, while still finding minimal-cost paths.