Unsupervised Learning: K-means Clustering and Anomaly Detection

Unsupervised Learning: K-means Clustering and Anomaly Detection

Explore core concepts of unsupervised learning, including K-means clustering, optimization strategies, and how anomaly detection systems are designed and evaluated.

Jongmin Lee
7 min read

1. Unsupervised learning

Clustering

What is clustering?

Screenshot 2024-04-09 at 5.51.38 PM.png

Screenshot 2024-04-09 at 5.52.01 PM.png

K-means intuition

Screenshot 2024-04-09 at 5.54.20 PM.png

Screenshot 2024-04-09 at 5.54.36 PM.png

Screenshot 2024-04-09 at 5.55.11 PM.png

Screenshot 2024-04-09 at 5.55.20 PM.png

Screenshot 2024-04-09 at 5.55.44 PM.png

Screenshot 2024-04-09 at 5.55.51 PM.png

K-means algorithm

Screenshot 2024-04-09 at 5.58.31 PM.png

Screenshot 2024-04-09 at 5.58.23 PM.png

Screenshot 2024-04-09 at 5.59.10 PM.png

Screenshot 2024-04-09 at 6.00.21 PM.png

Optimization objective

Screenshot 2024-04-09 at 6.05.08 PM.png

Screenshot 2024-04-09 at 6.05.18 PM.png

Screenshot 2024-04-09 at 6.05.38 PM.png

Screenshot 2024-04-09 at 6.05.49 PM.png

Initializing K-means

Screenshot 2024-04-09 at 6.09.01 PM.png

Screenshot 2024-04-09 at 6.10.12 PM.png

Screenshot 2024-04-09 at 6.13.20 PM.png

Screenshot 2024-04-09 at 6.15.46 PM.png

Choosing the number of clusters

Screenshot 2024-04-09 at 6.17.44 PM.png

Screenshot 2024-04-09 at 6.20.07 PM.png

Screenshot 2024-04-09 at 6.21.15 PM.png

Anomaly detection

Finding unusual events

Gaussian (normal) distribution

Anomaly detection algorithm

Developing and evaluating an anomaly detection system

Anomaly detection vs. supervised learning

Choosing what features to use