Eigenvectors, positive semi-definite matrices, and realizing your math is rusty.
Maximum Likelihood Estimation and priors. Statistics disguised as AI.
Least squares, Ridge, and Lasso. Penalize those large weights.
Sigmoid functions, cross-entropy loss, and Newton's Method.
Gaussian Discriminant Analysis, LDA, and QDA. Modeling the distributions.
Maximum margin classifiers, slack variables, and hinge loss.
Mapping data to infinite dimensions without actually doing the math.
Entropy, Information Gain, and greedily splitting your data.
Bagging, Random Forests, and Boosting (AdaBoost). Teamwork makes the dream work.
Perceptrons, activation functions, and the architecture of deep learning.
Taking derivatives using the chain rule across 50 layers of a network.
Principal Component Analysis, SVD, and dimensionality reduction.
K-Means, Gaussian Mixture Models, and EM algorithms. Finding patterns without labels.