Deep neural network image recognition method based on variable topological structure with variation particle swarm algorithm

Abstract

The invention discloses a deep neural network image recognition method based on a variable topological structure with a variation particle swarm algorithm. The method comprises the steps of preprocessing image data; searching the abstract dimension and the required feature number of the deep neural network for the features of the image data through a particle swarm algorithm; improving the exploration performance of the algorithm through mutation operation; optimizing the parameters of the deep neural network through a back propagation algorithm; and identifying the picture data to be identified. The advantages of high search speed and high efficiency of the particle swarm optimization algorithm are fully utilized; a particle swarm optimization deep neural network is used to optimize the abstract dimension and the required feature number of the features of the image data. The problem that the layer number and the node number of the deep network are determined only according to experiences of researchers in the prior art is solved, the performance of the deep neural network is further improved, and therefore the test time of the researchers is shortened, and the handwritten digit recognition accuracy is improved.

Description

technical field [0001] The invention belongs to the application field of computer analysis technology of image data, and particularly relates to a deep network topology optimization method based on particle swarm algorithm and a deep neural network image recognition method. Background technique [0002] The problem of image recognition is to use computer programs to process, analyze and understand the content in the picture, so that the computer can recognize various patterns of objects and objects from the picture. In recent years, image recognition has made many breakthroughs as an important field of artificial intelligence. [0003] Image classification is an important task in image processing. In the field of traditional machine learning, the standard process of identifying and classifying an image is feature extraction, feature screening, and finally inputting the feature vector into a suitable classifier to complete feature classification. Until 2006, Hinton first pr...

AI Technical Summary

Problems solved by technology

[0006] It can be seen that the selection of features affects the accuracy of image recognition to a large extent, but deep learning still has not fully automated feature engineering. The number of nodes corresponding to each depth of the network) is often determined by the experience of researchers, and there is no good way to automatically find a relatively good number of features

Images