A Kernel is a function used in algorithms like SVM to implicitly map data to higher-dimensional feature spaces. Kernel methods allow complex boundaries to be learned without explicit feature engineering. Common kernels include linear, polynomial, and RBF (Gaussian).