---
title: "Learning Canonical Correlations of Paired Tensor Sets Via Tensor-to-Vector Projection"
authors:
  - "Haiping Lu"
doi: ""
url: "https://ijcai.org/proceedings/2013/226"
pdf: "https://ijcai.org/proceedings/2013/0226.pdf"
venue: "Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence"
year: 2013
pages: "1516-1521"
track: ""
keywords:
---

# Learning Canonical Correlations of Paired Tensor Sets Via Tensor-to-Vector Projection

Haiping Lu

*Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence*, pages 1516-1521.

## Abstract

Canonical correlation analysis (CCA) is a useful technique for measuring relationship between two sets of vector data. For paired tensor data sets, we propose a multilinear CCA (MCCA) method. Unlike existing multilinear variations of CCA, MCCA extracts uncorrelated features under two architectures while maximizing paired correlations. Through a pair of tensor-to-vector projections, one architecture enforces zero-correlation within each set while the other enforces zero-correlation between different pairs of the two sets. We take a successive and iterative approach to solve the problem. Experiments on matching faces of different poses show that MCCA outperforms CCA and 2D-CCA, while using much fewer features. In addition, the fusion of two architectures leads to performance improvement, indicating complementary information.