Multi-view Generative Adversarial Networks

TL;DR

This paper introduces Multi-view BiGAN, a generative model that estimates data density from multiple views, handles missing views, and updates predictions as new views become available, advancing multi-view learning capabilities.

Contribution

It extends BiGAN to multi-view data, enabling density estimation, missing view handling, and dynamic prediction updates, which were not addressed together before.

Findings

MV-BiGAN performs effective density estimation from multi-view data.

The model can handle missing views during inference.

It updates predictions dynamically when new views are added.

Abstract

Learning over multi-view data is a challenging problem with strong practical applications. Most related studies focus on the classification point of view and assume that all the views are available at any time. We consider an extension of this framework in two directions. First, based on the BiGAN model, the Multi-view BiGAN (MV-BiGAN) is able to perform density estimation from multi-view inputs. Second, it can deal with missing views and is able to update its prediction when additional views are provided. We illustrate these properties on a set of experiments over different datasets.