Variational graph autoencoders for multiview canonical correlation analysis

作     者：Kaloga, Yacouba Borgnat, Pierre Chepuri, Sundeep Prabhakar Abry, Patrice Habrard, Amaury 

作者机构：Univ Lyon Univ Claude Bernard CNRS Ens LyonLab Phys Lyon France Indian Inst Sci Dept Elect & Commun Engn Bangalore Karnataka India Univ Lyon CNRS Lab Hubert Curien UJM St EtienneUMR 5516 F-5516 Lyon France 

出 版 物：《SIGNAL PROCESSING》 (信号处理)

年 卷 期：2021年第188卷

页      面：108182-108182页

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 

基　　金：IFCAM project [MA/IFCAM/19/56] ACADEMICS Grant of IDEXLYON, Univ. Lyon, PIA [ANR-16-IDEX-0005] ANR project DataRedux [ANR-19-CE46-0008] CBP IT test platform (ENS de Lyon, France) 

主　　题：Canonical correlation analysis Dimensionality reduction Multiview representation learning Graph neurals networks Variational inference 

摘      要：We present a novel approach for multiview canonical correlation analysis based on a variational graph neural network model. We propose a nonlinear model which takes into account the available graph based geometric constraints while being scalable to large-scale datasets with multiple views. This model combines the probabilistic interpretation of CCA with an autoencoder architecture based on graph convolutional neural network layers. Experiments with the proposed method are conducted on classification, clustering, and recommendation tasks on real datasets. The algorithm is competitive with state-of-the-art multiview representation learning techniques, in addition to being scalable and robust to instances with missing views. (c) 2021 Elsevier B.V. All rights reserved.