the kitchen 🍜

these posts serve as technical resources for learning and research

deep learning for computer vision, part 1: image classification and nearest neighbours

a scroll-through walk through the data-driven approach and k-nearest neighbours, on real CIFAR-10 photos of planes, cars, frogs and ships

13 min read · October 25, 2026

2026 · ai computer-vision deep-learning interactive · explainer
clustering from scratch, part 4: Gaussian mixtures and EM

clusters as overlapping, tilted Gaussian blobs, fitted by the EM algorithm; soft assignments, the log-likelihood that only goes up, and choosing k with BIC, all run live

14 min read · October 25, 2026

2026 · ml clustering unsupervised interactive · explainer
clustering from scratch, part 3: DBSCAN and density

clusters as crowded regions, grown point by point from core points, with a real notion of noise; DBSCAN run live, with ε and minPts on sliders

13 min read · October 25, 2026

2026 · ml clustering unsupervised interactive · explainer
clustering from scratch, part 2: hierarchical clustering

merge the two closest clusters, again and again, and read every possible clustering off one tree; with single, complete, average and Ward linkage run live

14 min read · October 25, 2026

2026 · ml clustering unsupervised interactive · explainer
clustering from scratch, part 1: k-means

a scroll-through, step-by-step walk through Lloyd's algorithm and k-means++, with every distance drawn and every number computed on the points you see

19 min read · October 25, 2026

2026 · ml clustering unsupervised interactive · explainer

deep learning for computer vision, part 1: image classification and nearest neighbours

clustering from scratch, part 4: Gaussian mixtures and EM

clustering from scratch, part 3: DBSCAN and density

clustering from scratch, part 2: hierarchical clustering

clustering from scratch, part 1: k-means