Examining Sigmoid vs ReLu Activation Functions in Deep Learning

Document Type

Contribution to Book

Publication Date


Publication Title

Interdisciplinary Research in Technology and Management


In recent years, deep learning has been considered to be a solution for many different problems such as natural language processing, pattern recognition, image detection and image classification. Artificial neural networks (ANN) are one of the deep learning models developed to address these problems. This study presents a Convolutional Neural Network (CNN) with LeNet architecture for image classification. Tests were conducted on the Caltech-101 datasets to determine the effectiveness of the CNN model. Over 1260 images were used and results indicate that the CNN with LeNEt was more accurate in image classification.


Georgia Southern University faculty members, Mohammad Anwarul Islam and Hayden Wimmer, co-authored Examining Sigmoid vs ReLu Activation Functions in Deep Learning.