Abstract:Multi-view representation learning is a promising and challenging research topic, which aims to integrate multiple data information from different views to improve the learning performance. The recent deep Gaussian processes (DGPs) have the advantages of good uncertainty estimates, powerful non-linear mapping ability and great generalization capability, which can be used as an excellent data representation learning method. However, DGPs only focus on single view data and are rarely applied to the multi-view scenario. In this paper, we propose a multi-view representation learning algorithm with deep Gaussian processes (named MvDGPs), which inherits the advantages of deep Gaussian processes and multi-view representation learning, and can learn more effective representation of multi-view data. The MvDGPs consist of two stages. The first stage is multi-view data representation learning, which is mainly used to learn more comprehensive representations of multi-view data. The second stage is classifier design, which aims to select an appropriate classifier to better employ the representations obtained in the first stage. In contrast with DGPs, MvDGPs support asymmetrical modeling depths for different views of data, resulting in better characterizations of the discrepancies among different views. Experimental results on real-world multi-view data sets verify the effectiveness of the proposed algorithm, which indicates that MvDGPs can integrate the complementary information in multiple views to discover a good representation of the data.

Multi-View Representation Learning with Manifold Smoothness

Multi-view Learning Via Multiple Graph Regularized Generative Model

Multi-View Similarity Learning Of Manifold Data

Geodesic Based Semi-supervised Multi-manifold Feature Extraction

Unsupervised Multi-Manifold Clustering by Learning Deep Representation

Multi-View Representation Learning with Deep Gaussian Processes

Multi-view low-rank matrix factorization using multiple manifold regularization.

Smooth Representation Learning from Multi-View Data.

Deep Manifold Regularized Semi-Nonnegative Matrix Factorization for Multi-view Clustering

Multi-View Representation Learning via Dual Optimal Transportation

Multi-view Multi-Manifold Learning with Local and Global Structure Preservation.

A Generative Model with Ensemble Manifold Regularization for Multi-view Clustering

General multi-view semi-supervised least squares support vector machines with multi-manifold regularization

Multi-View Graph Clustering by Adaptive Manifold Learning

Multi-view Manifold Learning with Locality Alignment

Learnable Graph Guided Deep Multi-view Representation Learning Via Information Bottleneck

Learning Multi-View Manifold for Single Image Based Modeling

Learning unsupervised node representation from multi-view network

Multi-Label Manifold Learning

Multi-View Clustering on Topological Manifold.

Multi-view Domain Generalization for Visual Recognition