Principal regression analysis
- 1 June 2011
- conference paper
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2881-2888
- https://doi.org/10.1109/cvpr.2011.5995618
Abstract
A new paradigm for multivariate regression is proposed; principal regression analysis (PRA). It entails learning a low dimensional subspace over sample-specific regressors. For a given input, the model predicts a subspace thought to contain the corresponding response. Using this subspace as a prior, the search space is considerably more constrained. An efficient local optimisation strategy is proposed for learning and a practical choice for its initialisation suggested. The utility of PRA is demonstrated on the task of non-rigid face and car alignment using challenging "in the wild" datasets, where substantial performance improvements are observed over alignment with a conventional prior.Keywords
This publication has 19 references indexed in Scilit:
- A globally optimal data-driven approach for image distortion estimationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2010
- Face alignment through subspace constrained mean-shiftsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2009
- Discriminative Face AlignmentIeee Transactions On Pattern Analysis and Machine Intelligence, 2008
- A Nonlinear Discriminative Approach to AAM FittingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2007
- On the Wiberg Algorithm for Matrix Factorization in the Presence of Missing ComponentsInternational Journal of Computer Vision, 2006
- Healing the relevance vector machine through augmentationPublished by Association for Computing Machinery (ACM) ,2005
- Orthogonal matching pursuit: recursive function approximation with applications to wavelet decompositionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Probabilistic visual learning for object representationIeee Transactions On Pattern Analysis and Machine Intelligence, 1997
- Updating Quasi-Newton Matrices with Limited StorageMathematics of Computation, 1980
- XVI. Functions of positive and negative type, and their connection the theory of integral equationsPhilosophical Transactions of the Royal Society A, 1909