series
Machine Learning Fundamentals
A planned series on the core ideas behind learning from data: models, loss functions, optimization, and how to evaluate what a model has learned.
This series is planned and has not started publishing. The outline below describes the intended scope and may change before any entry is written.
Entries
Learning from Data: Models, Loss Functions, and Risk
Framing prediction as minimizing a loss over examples, and what that does and does not tell us about unseen data.
Not written yet.
Gradient Descent and Its Variants
How iterative optimization works, why step size matters, and where stochastic variants fit in.
Not written yet.
Generalization, Overfitting, and Regularization
Why a model can fit its training data well and still predict poorly, and the standard ways of limiting that.
Not written yet.
Evaluating Models Honestly
Train, validation, and test splits; leakage; and how to report results together with their limitations.
Not written yet.