Enhancing the performance of CNN models for pneumonia and skin cancer detection using novel fractional activation function

Kumar, Meshach and Mehta, Utkal V. (2025) Enhancing the performance of CNN models for pneumonia and skin cancer detection using novel fractional activation function. Applied Soft Computing, 168 . NA. ISSN 1568-4946

Other (Fractional NN) - Published Version Restricted to Repository staff only Download (3MB)

Abstract

This paper introduces a novel Riemann–Liouville (RL) conformable fractional derivative based Adaptable-Shifted-Fractional-Rectified-Linear-Unit, briefly called RLASFReLU, and evaluates its efficacy in enhancing the performance of convolutional neural network (CNN) models for pneumonia and skin cancer detection. The study conducts a comprehensive comparative analysis against traditional activation functions and state-of-the-art CNN architectures. The results show that RLASFReLU consistently outperforms other functions, achieving higher accuracy. Comparative evaluations with various neural network architectures reveal that the model equipped with RLASFReLU exhibits superior performance despite its simplicity and fewer trainable parameters, highlighting its efficiency and effectiveness. The findings suggest that RLASFReLU holds promise in improving diagnostic accuracy and efficiency in medical imaging applications, contributing to advancements in healthcare technology and facilitating better patient care. The proposed fractional nonlinear transformation can offer high performance with reduced computational cost, making it practical for deployment in healthcare settings.

Item Type:	Journal Article
Uncontrolled Keywords:	Medical diagnosis; Image classification; CNN; Fractional calculus; Activation function
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > Robotics and Automation
Divisions:	School of Information Technology, Engineering, Mathematics and Physics (STEMP)
Depositing User:	Utkal Mehta
Date Deposited:	30 Jan 2025 00:07
Last Modified:	30 Jan 2025 00:07
URI:	https://repository.usp.ac.fj/id/eprint/14724

Actions (login required)

View Item