Evaluation of Visualization Methods' Effect on Convolutional Neural Networks Research
Haijing Tang, Yiru Wang, Xu Yang · 2018
In recent years, as an important research hotspot in the field of artificial intelligence and machine learning, the convolutional neural network has made substantial breakthroughs and has been widely used. In order to better explore and understand its structure, more and more researchers have shifted the focus of their research to the visualization of convolutional neural networks. They learned from the neural network what features were studied. They applied it to parameter adjustment and optimization in the convolutional neural networks and achieved good results. In this paper, the basic structure of convolutional neural network is described first. Secondly, some commonly used volume and neural network models are introduced. Finally, the convolutional neural network visualization technology is evaluated.