Incomplete Multi-View Clustering Method Based on Missing Graph Reconstruction and Adaptive Nearest Neighbors

Abstract

The invention discloses an incomplete multi-view clustering method based on missing graph reconstruction and self-adaptive neighbors. The method realizes the clustering of incomplete multi-view data by learning consistent non-negative features. The present invention considers the incomplete graph structure on different views and decomposes it into one view consistent feature and multiple view specific features, wherein the view consistent feature is used to preserve the neighbor graph structure information of multi-view data. The innovation of the present invention lies in rethinking the incomplete multi-view clustering problem from the perspective of graph structure decomposition of incomplete views, and learning the consistent non-negative features and common graph structures of missing multi-view data at the same time, wherein the consistent non-negative features satisfy the common Graph structure constraints. The incomplete multi-view clustering framework of the present invention is composed of a matrix decomposition model and an adaptive nearest neighbor model, and the training target and the derivation process are deduced at the same time; the method of the present invention can handle various incomplete multi-view scenarios without filling in missing views. .

Description

technical field [0001] The invention relates to the field of computer technology, to multi-view learning technology, and in particular to an incomplete multi-view clustering method based on missing graph reconstruction and self-adaptive neighbors. Background technique [0002] The background art involves three major blocks: a matrix decomposition algorithm based on non-negative and orthogonal constraints, a clustering algorithm based on adaptive neighbors, and a multi-view spectral clustering algorithm based on non-negative features and spectral features. [0003] 1) Matrix factorization algorithm based on non-negative and orthogonal constraints [0004] Spectral clustering is an algorithm that evolved from graph theory and was later widely used in clustering. Its main idea is to treat all data as points in space, and these points are connected by edges. Edges between two points that are farther apart have a lower weight value, while two points that are closer together hav...

AI Technical Summary

Problems solved by technology

[0033] Although the multi-view spectral clustering algorithm based on non-negative features and spectral features can obtain the consistent representation of multi-view data, it cannot learn the consistent graph structure of multi-view data.

That is to say, the algorithm cannot realize the joint learning of consistency graph structure and consistency features

Moreover, the multi-view spectral clustering algorithm based on non-negative features and spectral features can only handle complete multi-view data, and cannot handle missing view scenarios.

When any data is incomplete, the algorithm cannot learn the consistency characteristics of the missing data, so it cannot realize the clustering of the missing data.

Images