CaltechAUTHORS
  A Caltech Library Service

Analytic Continued Fractions for Regression: A Memetic Algorithm Approach

Moscato, Pablo and Sun, Haoyuan and Haque, Mohammad Nazmul (2021) Analytic Continued Fractions for Regression: A Memetic Algorithm Approach. Expert Systems with Applications, 179 . Art. No. 115018. ISSN 0957-4174. doi:10.1016/j.eswa.2021.115018. https://resolver.caltech.edu/CaltechAUTHORS:20210423-164854570

[img]
Preview
PDF - Accepted Version
See Usage Policy.

1MB

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20210423-164854570

Abstract

We present an approach for regression problems that employs analytic continued fractions as a novel representation. Comparative computational results using a memetic algorithm are reported in this work. Our experiments included fifteen other different machine learning approaches including five genetic programming methods for symbolic regression and ten machine learning methods. The comparison on training and test generalization was performed using 94 datasets of the Penn State Machine Learning Benchmark. The statistical tests showed that the generalization results using analytic continued fractions provide a powerful and interesting new alternative in the quest for compact and interpretable mathematical models for artificial intelligence.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1016/j.eswa.2021.115018DOIArticle
ORCID:
AuthorORCID
Moscato, Pablo0000-0003-2570-5966
Haque, Mohammad Nazmul0000-0002-0598-0867
Additional Information:© 2021 Elsevier. Received 11 February 2020, Accepted 6 April 2021, Available online 20 April 2021. This work was supported by The University of Newcastle, Caltech SURF, the Maitland Cancer Appeal and the Australian Government through the Australian Research Council’s Discovery Projects funding scheme (project DP200102364). We thank Dr Markus Wagner, School of Computer Science at The University of Adelaide, Australia for his thoughtful comments that helped us to improve an earlier version of the manuscript. We also thank the members of Prof. Jason H. Moore’s research lab at the University of Pennsylvania, USA, for making both the source code of their experiments and the Penn Machine Learning Benchmarks datasets available. M.N.H. and P.M. thank Renata Sarmet from the Universidade Federal de São Carlos, Sao Paulo, Brazil for discussion about the performance profile plot and sharing her Python code to produce one of the figures. CRediT authorship contribution statement: Pablo Moscato: Conceptualization, Methodology, Formal analysis, Investigation, Writing - original draft, Writing - review & editing, Supervision, Project administration, Funding acquisition. Haoyuan Sun: Methodology, Software, Validation, Formal analysis, Investigation, Writing - review & editing. Mohammad Nazmul Haque: Methodology, Software, Validation, Formal analysis, Investigation, Writing - original draft, Writing - review & editing, Visualization. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Funders:
Funding AgencyGrant Number
University of NewcastleUNSPECIFIED
Caltech Summer Undergraduate Research Fellowship (SURF)UNSPECIFIED
Maitland Cancer AppealUNSPECIFIED
Australian Research CouncilDP200102364
Subject Keywords:Symbolic regression; Memetic algorithm; Analytic continued fractions
DOI:10.1016/j.eswa.2021.115018
Record Number:CaltechAUTHORS:20210423-164854570
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20210423-164854570
Official Citation:Pablo Moscato, Haoyuan Sun, Mohammad Nazmul Haque, Analytic Continued Fractions for Regression: A Memetic Algorithm Approach, Expert Systems with Applications, Volume 179, 2021, 115018, ISSN 0957-4174, https://doi.org/10.1016/j.eswa.2021.115018. (https://www.sciencedirect.com/science/article/pii/S0957417421004590)
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:108828
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:28 Apr 2021 15:05
Last Modified:05 May 2021 19:42

Repository Staff Only: item control page