Multi-View Intact Space Learning
- 30 March 2015
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 37 (12), 2531-2544
- https://doi.org/10.1109/tpami.2015.2417578
Abstract
It is practical to assume that an individual view is unlikely to be sufficient for effective multi-view learning. Therefore, integration of multi-view information is both valuable and necessary. In this paper, we propose the Multi-view Intact Space Learning (MISL) algorithm, which integrates the encoded complementary information in multiple views to discover a latent intact representation of the data. Even though each view on its own is insufficient, we show theoretically that by combing multiple views we can obtain abundant information for latent intact space learning. Employing the Cauchy loss (a technique used in statistical learning) as the error measurement strengthens robustness to outliers. We propose a new definition of multi-view stability and then derive the generalization error bound based on multi-view stability and Rademacher complexity, and show that the complementarity between multiple views is beneficial for the stability and generalization. MISL is efficiently optimized using a novel Iteratively Reweight Residuals (IRR) technique, whose convergence is theoretically analyzed. Experiments on synthetic data and real-world datasets demonstrate that MISL is an effective and promising algorithm for practical applications.Keywords
Funding Information
- Australian Research Council Projects (FT-130101457, DP-140102164)
- NSFC (61375026, 2015BAF15B00, JCYJ 20120614152136201)
This publication has 28 references indexed in Scilit:
- Group Sparse Multiview Patch Alignment Framework With View Consistency for Image ClassificationIEEE Transactions on Image Processing, 2014
- Click Prediction for Web Image Reranking Using Multimodal Sparse CodingIEEE Transactions on Image Processing, 2014
- Large-Margin Multi-ViewInformation BottleneckIEEE Transactions on Pattern Analysis and Machine Intelligence, 2014
- Multiview Vector-Valued Manifold Regularization for Multilabel Image ClassificationIEEE Transactions on Neural Networks and Learning Systems, 2013
- Active co-analysis of a set of shapesACM Transactions on Graphics, 2012
- Concentration inequalities for dependent random variables via the martingale methodThe Annals of Probability, 2008
- The covering number in learning theoryJournal of Complexity, 2002
- Breakdown points of Cauchy regression-scale estimatorsStatistics & Probability Letters, 2002
- Combining labeled and unlabeled data with co-trainingPublished by Association for Computing Machinery (ACM) ,1998
- Linear convergence of generalized Weiszfeld's methodComputing, 1980