Scientific ComputingPaper
Face Recognition: PCA + SVM(May 2026)
The mathematics of face recognition: PCA whitening, eigenfaces, and kernel methods on LFW
Face recognition built from the math up, on the LFW benchmark: the point is the mathematics - PCA whitening and eigenfaces derived step by step, the kernel trick and Lagrange multipliers worked through, then implemented as a fully classical pipeline. The site pairs an interactive visualizer with a complete mathematical write-up and an animated slide deck rendered with Manim. Reproducible end to end: CLAHE preprocessing, dimensionality reduction, kernel selection, and evaluation are documented and implemented in the open repository. Presented at SCSS 2026, where it took first place among the master's presentations.
