Error PCA: A Density-Overlap Approach to Supervised Feature Extraction for Classification

• Cloves Cabral Vilela Filho UFAPE
• Anderson Pinheiro Cavalcanti UFRPE
• Tiago Buaque A. de Carvalho UFAPE

DOI: https://doi.org/10.5753/eniac.2026.31359

Resumo

A Análise de Componentes Principais (PCA — Principal Component Analysis) consolida-se como uma das técnicas mais difundidas para a extração de características. Este trabalho apresenta o EPCA (Error PCA), uma abordagem supervisionada que integra medidas de similaridade, especificamente, coeficientes de sobreposição de distribuições de probabilidade, ao PCA clássico. A validação experimental foi conduzida por meio da comparação da acurácia do EPCA com o PCA tradicional, o MCEPCA e o PCA Supervisionado (SPCA), avaliados sob quatro algoritmos de classificação em múltiplos conjuntos de dados. Os resultados demonstram que o método proposto igualou ou superou as técnicas comparadas em termos de acurácia preditiva.

Referências

Barshan, E., Ghodsi, A., Azimifar, Z., and Zolghadri Jahromi, M. (2011). Supervised principal component analysis: Visualization, classification and regression on subspaces and submanifolds. Pattern Recognition, 44(7):1357–1371.

Brownlee, J. (2020). Probability for Machine Learning.

chin Ivan Chang, Y. (2026). A survey: Potential dimensionality reduction methods. arXiv.

DasGupta, A. (2011). Probability for Statistics and Machine Learning. Springer Science.

de Carvalho, T. B. A., Sibaldo, M. A. A., Tsang, I. R., and Cavalcanti, G. D. d. C. (2017). Principal component analysis for supervised learning: a minimum classification error approach. Journal of Information and Data Management, 8(2):131–145.

Demšar, J. (2006). Statistical comparisons of classifiers over multiple data sets. Journal of Machine Learning Research, 7:1–30.

Dhaker, H., Ngom, P., Ibrahimou, B., and Mbodj, M. (2019). Overlap coefficients based on kullback-leibler of two normal densities: Equal means case. Journal of Mathematics Research.

Dua, D. and Graff, C. (2019). UCI machine learning repository. [link].

Fukunaga, K. (1990). Introduction to Statistical Pattern Recognition. Academic Press, San Diego, CA, EUA, 2 edition.

Inman, H. F. and Bradley Jr., E. L. (1989). The overlapping coefficient as a measure of agreement between probability distributions and point estimation of the overlap of two normal densities. Communications in Statistics – Theory and Methods, 18(10):3851–3874.

Jolliffe, I. T. (2002). Principal Component Analysis. Springer Series in Statistics. Springer, New York, 2 edition.

Jose, S., Thomas, S., and Mathew, T. (2019). Interval estimation of the overlapping coefficient of two exponential distributions. Journal of Statistical Theory and Applications, 18:26–32.

Linacre, J. M. (1996). Overlapping normal distributions. Rasch Measurement Transactions, 10(1).

MCALINN K., T. K. (2026). Optimal hold-out size in cross-validation. arXiv.

Mitchell, T. M. (1997). Machine Learning. McGraw-Hill.

Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., and Duchesnay, E. (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12:2825–2830.

Vogelstein, J.T., B. E. T. M. (2021). Supervised dimensionality reduction for big data. Nat Commun, 12.

Webb, A. R. (2002). Statistical Pattern Recognition. John Wiley & Sons, Chichester, Inglaterra, 2 edition.

Weitzman, M. S. (1970). Measures of overlap of income distributions of white and negro families in the united states. Technical Paper 22, U.S. Bureau of the Census, Washington, D.C.