Huber, M., & Raidl, G. (2022). Learning Beam Search: Utilizing Machine Learning to Guide Beam Search for Solving Combinatorial Optimization Problems. In Machine Learning, Optimization, and Data Science (pp. 283–298). Springer Nature Switzerland AG. https://doi.org/10.34726/3443
7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science (LOD 2021)
en
Event date:
4-Oct-2021 - 8-Oct-2021
-
Event place:
Grasmere, Lake District, United Kingdom of Great Britain and Northern Ireland (the)
-
Number of Pages:
16
-
Publisher:
Springer Nature Switzerland AG, Cham, Switzerland
-
Peer reviewed:
Yes
-
Keywords:
Beam search; Combinatorial optimization; Longest common subsequence problem; Machine learning
en
Abstract:
Beam search (BS) is a well-known incomplete breadth-first-search variant frequently used to find heuristic solutions to hard combinatorial optimization problems. Its key ingredient is a guidance heuristic that estimates the expected length (cost) to complete a partial solution. While this function is usually developed manually for a specific problem, we propose a more general Learning Beam Search (LBS) that uses a machine learning model for guidance. Learning is performed by utilizing principles of reinforcement learning: LBS generates training data on its own by performing nested BS calls on many representative randomly created problem instances. The general approach is tested on two specific problems, the longest common subsequence problem and the constrained variant thereof. Results on established sets of benchmark instances indicate that the BS with models trained via LBS is highly competitive. On many instances new so far best solutions could be obtained, making the approach a new state-of-the-art method for these problems and documenting the high potential of this general framework.
en
Project title:
Doktoratskolleg "Vienna Graduate School on Computational Optimization": W1260-N35 (Fonds zur Förderung der wissenschaftlichen Forschung (FWF))