AWDA: An Adaptive Wishart Discriminant Analysis
Linear Discriminant Analysis (LDA) is widely-used for supervised dimension reduction and linear classification. Classical LDA, however, suffers from the ill-posed estimation problem on data with high dimension and low sample size (HDLSS). To cope with this problem, in this paper, we propose an Adaptive Wishart Discriminant Analysis (AWDA) for classification, that makes predictions in an ensemble way. Comparing to existing approaches, AWDA has two advantages: 1) leveraging the Wishart distribution, AWDA ensembles multiple LDA classifiers parameterized by the sampled covariance matrices via a Bayesian Voting Scheme, which theoretically improves the robustness of classification, compared to LDA classifiers using a single (probably ill-posed) covariance matrix estimator; 2) AWDA updates the weights for voting optimally to adapt the local information of each new input data, so as to enable the nonlinear classification. Theoretical analysis indicates that AWDA guarantees a close approximation to the optimal Bayesian inference and thus achieves robust performance on high dimensional data. Extensive experiments on real-world datasets show that our approach outperforms state-of-the-art algorithms by a large margin.
H. Xiong et al., "AWDA: An Adaptive Wishart Discriminant Analysis," Proceedings of the 17th IEEE International Conference on Data Mining (2017, New Orleans, LA), Institute of Electrical and Electronics Engineers (IEEE), Nov 2017.
The definitive version is available at https://doi.org/10.1109/ICDM.2017.62
17th IEEE International Conference on Data Mining, ICDM (2017: Nov. 18-21, New Orleans, LA)
Mathematics and Statistics
Keywords and Phrases
Data Mining; Classification; Linear Discriminant Analysis; Bayesian Inference and Wishart Distribution
International Standard Book Number (ISBN)
International Standard Serial Number (ISSN)
Article - Conference proceedings
© 2017 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.