Image classification method based on reliable weight optimal transmission

Abstract

The invention discloses an image classification method based on reliable weight optimal transmission, and the method comprises the following steps: firstly carrying out the preprocessing of source domain data, and enabling a deep neural network to fit a sample label of a source domain sample image; marking a picture, marking a pseudo label on the target domain data sample, pairing nodes to realizepairing of associated pictures in a source domain and a target domain, and finally realizing automatic analysis through a feature extractor and a self-adaptive discriminator to classify the images. The invention provides a subspace reliability method for dynamically measuring sample inter-domain differences by utilizing space prototype information and an intra-domain structure. The method can beused as a pretreatment step of an adaptive technology in the prior art, and the efficiency is greatly improved. According to the method, the reliability of the contraction subspace is combined with the optimal transportation strategy, so that the depth characteristics are more obvious, and the robustness and effectiveness of the model are enhanced. The deep neural network works stably on various data sets, and the performance of the deep neural network is superior to that of an existing method.

Description

technical field [0001] The invention relates to the field of image classification, in particular to an image classification method based on reliable weight optimal transmission. Background technique [0002] Deep learning is an important method in the field of computer vision. It learns the internal laws and representation levels of sample data through training, and is widely used in image classification, object detection, and semantic segmentation. Traditional supervised learning requires a large amount of manually labeled data, which is very time-consuming and laborious. In order to avoid mechanically repetitive labeling work, the Unsupervised Domain Adaptation (UDA) method aims to apply the knowledge or patterns learned in a certain field to new and different but related fields, with rich supervision. The source domain of the information (Source Domain) to improve the performance of the target domain (Target Domain) model with no labels or only a few labels. Among them,...

AI Technical Summary

Problems solved by technology

The difference between the two domain distributions is the key to unsupervised domain adaptive technology, but when describing this difference, the existing research often ignores the prototype information and the structural information in the domain, resulting in the lack of mining of latent semantic information.

[0025] 2. Negative transfer

[0026] In the optimal transmission process of the existing technology, due to the dissimilarity between the source domain and the target domain, or because the transfer learning method does not find transferable components, it may cause the knowledge learned on the source domain to be ineffective for the learning on the target domain. Negative effect, that is, negative transfer (negative transfer)

[0027] 3. Clustering features are not significant enough

The deep clustering features mined by existing technologies are not significant enough, and the robustness and effect are not good enough

Images