What question did this study set out to answer?

The aim is to automate the development of algorithms for NP-hard combinatorial optimization problems using learned heuristics.

April 5, 2017Open Access

Learning Combinatorial Optimization Algorithms over Graphs

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial explores heuristic algorithm learning in optimization problems over graphs, suggesting automated solutions.

The aim is to automate the development of algorithms for NP-hard combinatorial optimization problems using learned heuristics.
Proposed a novel approach combining reinforcement learning and graph embedding to learn algorithms.
Developed a greedy policy that constructs solutions incrementally based on graph embedding outputs.
Applied the framework to optimization problems such as Minimum Vertex Cover, Maximum Cut, and Traveling Salesman.
The framework effectively learns algorithms that perform well on various graph-based optimization problems.
Demonstrated improved algorithmic performance on NP-hard problems compared to traditional methods.

Dai et al. (2017) studied this question.

View Full Paper

Ask AI

Bookmark