A discrete learning-based intelligent classifier for breast cancer classification

Precise diagnosis of benign and malignant breast cancer plays an important role in the effective treatment of breast cancer patients. Several classification models with different characteristics have been developed and used in a wide range of breast cancer domains to improve classification accuracy....

Full description

Saved in:
Bibliographic Details
Published in:Multimedia tools and applications Vol. 83; no. 32; pp. 78269 - 78292
Main Authors: Khashei, Mehdi, Bakhtiarvand, Negar, Ahmadi, Parsa
Format: Journal Article
Language:English
Published: New York Springer US 01.09.2024
Springer Nature B.V
Subjects:
Breast cancer
Classification
Computer Communication Networks
Computer Science
Continuity (mathematics)
Cost function
Data Structures and Information Theory
Learning
Multilayer perceptrons
Multimedia Information Systems
Special Purpose and Application-Based Systems
Track 3: Biometrics and HCI
Breast Cancer datasets
Discrete and Continuous learning algorithms
Breast Cancer (BC)
Multilayer perceptron (MLP)
Classification
ISSN:1573-7721, 1380-7501, 1573-7721
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Precise diagnosis of benign and malignant breast cancer plays an important role in the effective treatment of breast cancer patients. Several classification models with different characteristics have been developed and used in a wide range of breast cancer domains to improve classification accuracy. Although the classification models differ in different aspects, they all have the same logic in their learning processes and use a continuous distance-based cost function. However, using a continuous distance-based function as a cost function in the learning processes of the traditional classification models is unreasonable or at least insufficient; since the goal function of the classification, is discrete. Hence, developing a discrete cost function for learning the classification problems, due to more consistency, may improve the classification rate; but, it has been neglected in the literature. In this paper, in contrast to all traditional continuous distance-based learning processes, a novel discrete learning-based process is proposed and implemented on a multilayer perceptron to yield a more consistent intelligent classifier. Then, the proposed discrete learning-based multilayer perceptron (DIMLP) is used for breast cancer classification. Empirical results of the breast cancer datasets indicate that the proposed DIMLP model can averagely achieve the classification rate of 94.70%, while the classification rate for the traditional MLP model is only equal to 88.54%. Therefore, the proposed DIMLP can be an appropriate and efficient alternative model for intelligent breast cancer classification, especially when more accurate results and/or a more reasonable model are required.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1573-7721
1380-7501
1573-7721
DOI:10.1007/s11042-024-18646-6