Title
Learning Greedy Algorithms for Maximum Independent Set
Presentation Type
Event
Start Date
27-4-2019 10:50 AM
End Date
27-4-2019 11:29 AM
Abstract
In 2017, Dai, Khalil, Zhang, Dilkina, and Song introduced a machine learning framework for finding greedy algorithm heuristics for many common combinatorial optimization problems. In particular, they implemented their framework to create greedy algorithms for the minimum vertex cover, maximum cut and the traveling salesman problem.
We use this framework to study the problem of finding large independent sets in random regular graphs and random graphs with planted independent sets. We compare the sizes of independent sets found by the learned algorithms to those found by simple heuristics such as (random) GREEDY and MINGREEDY.
Learning Greedy Algorithms for Maximum Independent Set
In 2017, Dai, Khalil, Zhang, Dilkina, and Song introduced a machine learning framework for finding greedy algorithm heuristics for many common combinatorial optimization problems. In particular, they implemented their framework to create greedy algorithms for the minimum vertex cover, maximum cut and the traveling salesman problem.
We use this framework to study the problem of finding large independent sets in random regular graphs and random graphs with planted independent sets. We compare the sizes of independent sets found by the learned algorithms to those found by simple heuristics such as (random) GREEDY and MINGREEDY.