03 — Kernel Methods

Methods that operate via kernel functions to learn non-linear decision boundaries in high-dimensional implicit feature spaces. SVMs are the canonical example.

Notes

• Kernel Methods and SVMs — hard/soft-margin SVM, kernel trick, RBF kernel, SVR, kernel ridge regression
• Kernel Methods — Implementation — scikit-learn SVC/SVR, kernel selection, hyperparameter tuning, decision boundary visualisation

Links

← 02 — Tree-Based Models → 04 — Instance-Based Methods

• Lagrangian and Constrained Optimization
• Linear Algebra — kernel inner products