Cross-view kernel transfer

TL;DR

This paper introduces Cross-View Kernel Transfer (CVKT), a method for completing kernel matrices in multi-view data by learning transformations between views, demonstrated on simulated, biological, and gesture datasets.

Contribution

The paper proposes CVKT, a novel approach for kernel matrix completion in multi-view data using learned transformations via kernel alignment.

Findings

Effective in completing kernel matrices with missing data across views.

Improves classification accuracy in multi-view datasets.

Applicable to biological and gesture recognition datasets.

Abstract

We consider the kernel completion problem with the presence of multiple views in the data. In this context the data samples can be fully missing in some views, creating missing columns and rows to the kernel matrices that are calculated individually for each view. We propose to solve the problem of completing the kernel matrices with Cross-View Kernel Transfer (CVKT) procedure, in which the features of the other views are transformed to represent the view under consideration. The transformations are learned with kernel alignment to the known part of the kernel matrix, allowing for finding generalizable structures in the kernel matrix under completion. Its missing values can then be predicted with the data available in other views. We illustrate the benefits of our approach with simulated data, multivariate digits dataset and multi-view dataset on gesture classification, as well as with real biological datasets from studies of pattern formation in early \textit{Drosophila melanogaster} embryogenesis.