classical-mlsupervisedclassificationkernels
Support vector machines
Maximum-margin classifiers with the kernel trick — a clean lens on generalisation and duality.
Уровни глубины
L0Intro~2ч
Understands "widest street" geometric intuition on a 2D dataset.
L1Basics~12ч
Writes primal and dual QP; computes margin, support vectors; uses RBF kernel.
L2Working~18ч
Tunes C, gamma via grid + CV; uses sklearn SVC on medium data; applies kernel trick creatively.
L3Advanced~30ч
Derives SMO, structured/one-class SVM; analyses Mercer kernels and RKHS.
L4Research~60ч
Contributes to kernel learning, large-scale SVM, or connection to neural tangent kernel.
Ресурсы
L1 — Basics
L2 — Working