Research On The Multi-label Lassification Methods With The Label Embedding And Structure Information

Multi-label learning,an important research topic in the field of machine learning,has attracted widespread attention of scholars.50 far,though a series of efficient multi-label classification algorithms have been proposed,more emphasis should be put on the multi-label learming in complex situations,such as missing labels,excessive labels,and label noise.This thesis mostly focuses on coping with the challenges of multi-label classification under the circumstance of overmuch label catergories and label noise?by researching on how to effectively uncover and utilize the relationship among label correlations,feature space and label space.The main research work is as follows.1.We propose a novel deep neural network(DNN)based model,namely Deep Correlation Structure Preserved Label Space Embedding(DCSPE).Specifically,DCSPE derives a deep latent space by performing feature-aware label space embedding with deep canonical correlation analysis(DCCA)and preserving the intrinsic structure of the previous label space with proposed deep multidimensional scaling(DMDS).Our DCSPE is achieved by integrating the DNN architectures of the two DNN based models and ean leam a feature-aware structure preserved deep latent space.Furthermore,extensive experimental results on datasets with many labels demonstrate that our proposed approach is significantly better than the existing label embedding algorithms.2.We propose a novel deep neural network(DNN)based model for learning an effective deep latent space,namely Deep Cross-view label spaee Embedding with Correlation and Structure preserved(DCECS).In DCECS,the latent space correlates with feature and label spaces closely by virtue of the deep cross-view embedding.Meanwhile,the latent space is also learned under the guidance of label correlation and local structure of feature space which are exploited by hypergraph and graph regularizations.The overall framework achieves the complementarity and correspondence between information of feature and label space,therefore the feature-aware deep latent space we learmed has strong predictability and discriminant ability.Extensive experimental results on datasets with many labels demonstrate that our proposed approach is significantly better than the existing label embedding algorithms.3.We propose a novel cross-view based model which performs a robust and discriminant embedding,namely Robust Cross-view Embedding with Discriminant Structure for multi-label classification(RCEDS).In RCEDS,a novel hypergraph fusion technology is designed to explore and utilize the complementary between feature space and label space to make the proposed RCEDS robust.Meanwhile,we use double-side metric learning to mine the consistency between feature space and label space which effectively improve the discriminative ability of our proposed RCEDS.Furthermore,we conduct a deep extension for RCEDS,which is effectively applied to image annotation.Extensive experimental results on datasets with many labels demonstrate that our proposed approach is significantly better than the existing label embedding algorithms.