8. Machine learning foundations

Frame supervised, unsupervised, and reinforcement learning problems; create train, validation, and test splits; and recognize underfitting, overfitting, leakage, and distribution shift. Build classical baselines to judge whether deep learning is justified.