Semi-supervised generative adversarial network image classification method based on local manifold regularization

Abstract

The invention discloses a semi-supervised generative adversarial network image classification method based on local manifold regularization. According to the method, local manifold regularization is introduced into a semi-supervised generative adversarial network. According to the method, local manifold regularization is introduced based on the excellent fitting capability of the generative adversarial network to the data manifold, so that the problem of excessive training of the discriminator can be well solved. The manifold regularization item is added into the loss functions of the discriminator and the generator to punify the sudden change of the data manifold, so that the model can be prevented from falling into local collapse, the local disturbance of the model to the data manifold is enhanced to keep invariance, and the model has better robustness. The semi-supervised generative adversarial network image classification method combined with local manifold regularization can significantly improve the accuracy of image classification in the aspect of semi-supervised image classification.

Description

technical field [0001] The invention belongs to the field of image semi-supervised classification, in particular to an image classification method based on local manifold regularization semi-supervised generation confrontation network. . Background technique [0002] Semi-supervised learning (SSL) is a learning method that combines supervised learning and unsupervised learning in the field of machine learning. The semi-supervised classification method is to use a small amount of labeled data and a large amount of unlabeled data to train the model to optimize the performance of the classifier. At present, the commonly used semi-supervised learning methods are: semi-supervised support vector machine, collaborative training, self-training, graph theory semi-supervised learning, regularization constraint class and semi-supervised generative confrontation network, etc. Among them, Semi-supervised Learning with Generative Adversarial Networks (Semi-supervised Learning with Gener...

AI Technical Summary

Problems solved by technology

[0004] The problem to be solved by the present invention is that, in the semi-supervised learning of Feture Matching GAN, no regularization method is used, which may cause the model to fall into local collapse

Images