An Analysis of Convolutional Neural Network for Fashion Images Classification (Fashion-MNIST)

摘要

Recently, Convolutional Neural Networks (CNN) has been used in variety of domains, including fashion classification. Social media, e-commerce, and criminal law are extensively applicable in this field. CNNs are efficient to train and found to give the most accurate results in solving real world problems. In this paper, we use Fashion MNIST dataset for evaluating the performance of convolutional neural network based deep learning architectures. We compare most common deep learning architectures such as AlexNet, GoogleNet, VGG, ResNet, DenseNet and SqueezeNet to find the best performance. We additionally propose a simple modification to the architecture to improve and accelerate learning process. We report accuracy measurements (93.43%) and the value of loss function (0.19) using our proposed method and show its significant improvements over other architectures.