Probabilistic Fisher discriminant analysis

Fisher Discriminant Analysis (FDA) is a powerful and popular method for dimensionality reduction and classification which has unfortunately poor performances in the cases of label noise and sparse labeled data. To overcome these limitations, we propose a probabilistic framework for FDA and extend it to the semi-supervised case. Experiments on realworld datasets show that the proposed approach works as well as FDA in standard situations and outperforms it in the label noise and sparse label cases.