Matlab source codes of the Probabilistic Rank-One Discriminant Analysis (PRODA) algorithm presented in the paper Probabilistic Rank-One Discriminant Analysis via Collective and Individual Variation Modeling.
Face recognition with PRODA on 2D images from the FERET dataset:
Demo_PRODA.m
- DBpart.mat stores the indices for training (2 samples per class)/test data partition.
- FERETC80A45.mat stores 320 faces (32x32) of 80 subjects (4 samples per class) from the FERET dataset.
- Demo_PRODA.m provides example usage of PRODA for subspace learning and classification on 2D facial images.
- PRODA.m implements the PRODA algorithm described in the paper.
- projPRODA.m projects 2D data into the subspace learned by PRODA.
- sortProj.m sorts features by their Fisher scores in descending order.
- logdet.m computes the logarithm of determinant of a matrix.
If you find our codes helpful, please consider cite the following paper:
@article{
zhou2020PRODA,
title={Probabilistic Rank-One Discriminant Analysis via Collective and Individual Variation Modeling},
author={Yang Zhou and Yiu-ming Cheung},
journal={IEEE Transactions on Cybernetics},
year={2020},
volume={50},
number={2},
pages={627-639},
doi={10.1109/TCYB.2018.2870440},
}