A Thorough Review on Deep Learning Neural Network

Abstract

Deep Neural Network Topologies are used by subfield of machine learning called "deep learning" that are similar but different to handle a variety of challenges in domains including bioinformatics, computer vision and, among others. Recent research on deep learning has grown significantly across several applications. Deep learning technique produces state-of-the-art results by using numerous layers of features or data representations. Deep learning is essentially the application of neural networks with multiple hidden layers of neurons. In particular, this review paper firstly aims to offer a more thorough overview of the most fundamental deep learning components, further shows how deep learning techniques outperformed well-known ML techniques and then outlines how to deploy and build deep learning model. Secondly, Convolutional neural networks (CNN), most common used deep learning networks, are then introduced, along with a description of how they have implemented with matrix representation. Thirdly, we concentrate on application domain of deep learning, with an emphasis on its use in object detection. Finally, Future study directions are provided after the conclusion of the publication to assist scholars in understanding the research gaps and findings.