A0472
Title: Dynamic supervised principal component analysis for classification
Authors: Wenbo Ouyang - University of Arizona (United States)
Ruiyang Wu - Baruch College, CUNY (United States) [presenting]
Ning Hao - University of Arizona (United States)
Hao Zhang - University of Arizona (United States)
Abstract: A novel framework is introduced for high-dimensional dynamic classification to address the evolving nature of class distributions over time or other index variables. Under this framework, traditional discriminant analysis techniques are adapted to learn dynamic decision rules with respect to the index variable. It features a new supervised dimension reduction method employing kernel smoothing to identify the optimal subspace projection, and the resulting variables from this projection are trained with a subsequent classifier, such as linear discriminant analysis and quadratic discriminant analysis. The effectiveness of the proposed methods is illustrated through theoretical analysis and numerical examples. The results show considerable improvements in classification accuracy and computational efficiency. The contribution is to the field by offering a robust and adaptive solution to the challenges of scalability and non-staticity in high-dimensional data classification.
