A comparative study of the leading machine learning techniques and two new optimization algorithms

•Comparison of fourteen machine learning algorithms on a diverse collection of data sets.•First thorough analysis of combinatorial optimization algorithms for machine learning.•Combinatorial optimization algorithms achieve best and most robust performance.•All pairwise-similarities-based algorithms...

Full description

Saved in:
Bibliographic Details
Published in:European journal of operational research Vol. 272; no. 3; pp. 1041 - 1057
Main Authors: Baumann, P., Hochbaum, D.S., Yang, Y.T.
Format: Journal Article
Language:English
Published: Elsevier B.V 01.02.2019
Subjects:
Binary classification
Comparative study
Data mining
Supervised machine learning
Supervised normalized cut
Supervised machine learning
Data mining
Binary classification
Comparative study
Supervised normalized cut
ISSN:0377-2217, 1872-6860
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Comparison of fourteen machine learning algorithms on a diverse collection of data sets.•First thorough analysis of combinatorial optimization algorithms for machine learning.•Combinatorial optimization algorithms achieve best and most robust performance.•All pairwise-similarities-based algorithms are top performers. We present here a computational study comparing the performance of leading machine learning techniques to that of recently developed graph-based combinatorial optimization algorithms (SNC and KSNC). The surprising result of this study is that SNC and KSNC consistently show the best or close to best performance in terms of their F1-scores, accuracy, and recall. Furthermore, the performance of SNC and KSNC is considerably more robust than that of the other algorithms; the others may perform well on average but tend to vary greatly across data sets. This demonstrates that combinatorial optimization techniques can be competitive as compared to state-of-the-art machine learning techniques. The code developed for SNC and KSNC is publicly available.
ISSN:0377-2217
1872-6860
DOI:10.1016/j.ejor.2018.07.009