Unsupervised Learning

Methods for learning structure from unlabelled data: clustering, dimensionality reduction, density estimation, and anomaly detection.

Notes

• Clustering — K-Means (Lloyd’s, k-means++, silhouette), DBSCAN, HDBSCAN, hierarchical clustering, GMM soft clustering
• Clustering Implementation — K-Means elbow/silhouette selection, DBSCAN with k-distance graph, agglomerative with dendrogram, GMM BIC selection

Links

← Kernel Methods → Transformers

• Eigenvalues and Eigenvectors (PCA)
• Probabilistic Models (GMM)