Improvement in Convolutional Neural Network for CIFAR-10 Dataset Image Classification

Image classification requires the generation of features capable of detecting image patterns informative of group identity. The objective of this study was to classify images from the public CIFAR10 image dataset by leveraging combinations of disparate image feature sources from deep learning approaches. The majority of regular convolutional neural networks (CNN) are based on the same structure: modification of convolution and the process of max-pooling layers connected with a number of entirely linked layers. In this paper, the prime objective is to improve the effectiveness of simple convolutional neural network models. The Artificial Neural Network (ANN) algorithm is applied on a Canadian Institute For Advanced Research dataset (CIFAR-10) using two different CNN structures. The result of the improved model achieves 88% classification accuracy rate by running for 10 hours. The deep learning models are implemented with the use of Keras library available for Python programming language.