Research On Multi-view Subspace Learning And Application

Nowadays,alone with the development of information technology,the internet has burst out with a large number of diversified data with rich information.With the increasing amount of data and increasing complexity,low-dimensional subspace learning has become a research hotspot in the field of machine learning and pattern recognition.By means of mapping the data from the original high-dimensional manifold to the low-dimensional subspace,the problems in the high-dimensional space can be solved.And we can obtain ideal learning effect of the subspace learning model.With the development of feature extraction technique,one sample always can be represented by multiple features which locate in high-dimensional space.Multiple features can reflect various perspectives of one same sample,so there must be compatible and complementary information among the multiple views.Therefore,it becomes natural for one to integrate them together to obtain better performance rather than rely on just one single view.We consider effectively exploring and exploiting multiple representations simultaneously,so the key of multi-view learning is to leverage the complementary information from multiple views,which is of vital importance but challenging.However,most multi-view dimension reduction methods cannot handle multiple features from nonlinear space with high dimensions.To address this problem,we propose a novel multi-view dimension reduction method named Multi-view Reconstructive Preserving Embedding(MRPE)in this paper.MRPE reconstructs each sample by utilizing its k neighbors.The similarities between each sample and its neighbors is primely mapped into lower-dimensional space in order to preserve the underlying neighborhood structure of the original manifold.MRPE fully exploits correlations between each sample and its neighbors from multiple views by linear reconstruction.Furthermore,MRPE constructs an optimal problem and derives an iterative procedure to obtain the low-dimensional embedding.Various evaluations based on the applications of document classification,face recognition and image retrieval demonstrate the effectiveness of our proposed approach on multi-view dimension reduction.