Explore the Performance of Capsule Neural Network Learning Discrete Features

Abstract

The continuous breakthrough of deep learning model in image processing, natural language processing and other fields is mainly due to the strong ability of deep neural network in feature extraction. Based on the idea of capsule neural network, this paper proposes a capsule neural network for general classification problems, and explores the learning ability of capsule network model for classification problems of discrete feature. In order to evaluate the capsule network model, this paper verifies the effect of the model on real datasets, and makes a comparative analysis with common machine learning classification algorithms.