Binary Bleed: Fast Distributed and Parallel Method for Automatic Model Selection

dc.contributor.authorBarron, Ryan
dc.contributor.authorEren, Maksim
dc.contributor.authorBhattarai, Manish
dc.contributor.authorBoureima, Ismael
dc.contributor.authorMatuszek, Cynthia
dc.contributor.authorAlexandrov, Boian S.
dc.date.accessioned2024-08-20T13:45:33Z
dc.date.available2024-08-20T13:45:33Z
dc.date.issued2024-07-26
dc.description28th Annual IEEE High Performance Extreme Computing Virtual Conference, 23 - 27 September 2024
dc.description.abstractIn several Machine Learning (ML) clustering and dimensionality reduction approaches, such as non-negative matrix factorization (NMF), RESCAL, and K-Means clustering, users must select a hyper-parameter k to define the number of clusters or components that yield an ideal separation of samples or clean clusters. This selection, while difficult, is crucial to avoid overfitting or underfitting the data. Several ML applications use scoring methods (e.g., Silhouette and Davies Boulding scores) to evaluate the cluster pattern stability for a specific k. The score is calculated for different trials over a range of k, and the ideal k is heuristically selected as the value before the model starts overfitting, indicated by a drop or increase in the score resembling an elbow curve plot. While the grid-search method can be used to accurately find a good k value, visiting a range of k can become time-consuming and computationally resource-intensive. In this paper, we introduce the Binary Bleed method based on binary search, which significantly reduces the k search space for these grid-search ML algorithms by truncating the target k values from the search space using a heuristic with thresholding over the scores. Binary Bleed is designed to work with single-node serial, single-node multi-processing, and distributed computing resources. In our experiments, we demonstrate the reduced search space gain over a naive sequential search of the ideal k and the accuracy of the Binary Bleed in identifying the correct k for NMFk, K-Means pyDNMFk, and pyDRESCALk with Silhouette and Davies Boulding scores. We make our implementation of Binary Bleed for the NMF algorithm available on GitHub.
dc.description.sponsorshipThis research used resources provided by the LANL Institutional Computing Program supported by the U.S. Department of Energy National Nuclear Security Administration under Contract No. 89233218CNA000001.
dc.description.urihttp://arxiv.org/abs/2407.19125
dc.format.extent8 pages
dc.genreconference papers and proceedings
dc.genrepreprints
dc.identifierdoi:10.13016/m2yzl8-yxpg
dc.identifier.urihttps://doi.org/10.48550/arXiv.2407.19125
dc.identifier.urihttp://hdl.handle.net/11603/35728
dc.language.isoen_US
dc.relation.isAvailableAtThe University of Maryland, Baltimore County (UMBC)
dc.relation.ispartofUMBC Computer Science and Electrical Engineering Department
dc.relation.ispartofUMBC Student Collection
dc.relation.ispartofUMBC Faculty Collection
dc.rightsThis work was written as part of one of the author's official duties as an Employee of the United States Government and is therefore a work of the United States Government. In accordance with 17 U.S.C. 105, no copyright protection is available for such works under U.S. Law.
dc.rightsPublic Domain
dc.rights.urihttps://creativecommons.org/publicdomain/mark/1.0/
dc.subjectComputer Science - Artificial Intelligence Computer Science - Distributed, Parallel, and Cluster Computing Computer Science - Performance
dc.titleBinary Bleed: Fast Distributed and Parallel Method for Automatic Model Selection
dc.typeText
dcterms.creatorhttps://orcid.org/0009-0005-5045-9527
dcterms.creatorhttps://orcid.org/0000-0002-4362-0256
dcterms.creatorhttps://orcid.org/0000-0003-1383-8120

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
2407.19125v1.pdf
Size:
783.4 KB
Format:
Adobe Portable Document Format