A comparative performance of different convolutional neural network activation functions on image classification

Azhary, Muhammad Zulhazmi Rafiqi and Ismail, Amelia Ritahani (2024) A comparative performance of different convolutional neural network activation functions on image classification. International Journal on Perceptive and Cognitive Computing (IJPCC), 10 (2). pp. 118-122. E-ISSN 2462-229X

PDF - Published Version Restricted to Repository staff only Download (210kB) | Request a copy

Abstract

Activation functions are crucial in optimising Convolutional Neural Networks (CNNs) for image classification. While CNNs excel at capturingspatial hierarchies in images, the activation functions substantially impact their effectiveness. Traditional functions, such as ReLU and Sigmoid, have drawbacks, including the "dying ReLU" problem and vanishing gradients, which can inhibit learning and efficacy. The study seeks to comprehensively analyse various activation functions across different CNN architectures to determine their impact on performance. The findings suggest that Swish and Leaky ReLU outperform other functions, with Swish particularly promising in complicated networks such as ResNet. This emphasises the relevance of activation function selection in improving CNN performance and implies that investigating alternative functions can lead to more accurate and efficient models for image classification tasks.

Item Type:	Article (Journal)
Uncontrolled Keywords:	Activation Functions, Convolutional Neural Network , Image Classification
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Information and Communication Technology Kulliyyah of Information and Communication Technology Kulliyyah of Information and Communication Technology > Department of Computer Science Kulliyyah of Information and Communication Technology > Department of Computer Science
Depositing User:	Amelia Ritahani Ismail
Date Deposited:	17 Dec 2024 16:35
Last Modified:	17 Dec 2024 16:35
URI:	http://irep.iium.edu.my/id/eprint/116734

Actions (login required)

View Item