TransferLab Training: Practical Anomaly Detection - Module 6: Extreme Value Theory for AD

In this module, we introduce extreme value theory (EVT) as a method for finding thresholds for anomaly detection. EVT is based on the assumption that the scores of anomalous points are significantly higher than the scores of nominal points. It estimates the tail of the score distribution and uses this to find a probabilistically interpretable threshold. We apply EVT to the New York taxi dataset to find a threshold for detecting anomalies in the number of taxi rides.

Accessing the resources of this course:
🔗 GitHub Repository: complete code and examples on our GitHub https://github.com/aai-institute/tfl-...
🌐 Website: digests of the latest research on the TransferLab’s website: https://transferlab.ai/
🎓 Full Course & Certification (For FREE): Enroll in our full video course available on our learning platform.
Complete the course at your own pace and earn a certificate for free to enhance your portfolio: https://lms.appliedai-institute.de/

00:50 Extreme Value Theory (EVT)
09:30 Exercise: Playing with the GEV Parameters, Loading Data and Setting up Block Maxima
11:30 Solution: Playing with the GEV Parameters, Loading Data and Setting up Block Maxima
17:48 Exercise: MLE for the GEV
18:12 Solution: MLE for the GEV
26:26 Exercise: Improving the MLE fit
27:25 Q-Q Plot
28:52 Solution: Improving the MLE fit
33:51 Exercise: Uncertainty Estimation in GEV Distributions
35:39 Solution: Uncertainty Estimation in GEV Distributions
37:27 Exercise: GEV Distributions for Minima
38:19 Solution: GEV Distributions for Minima
40:01 Example: Comparison with the Z-test
41:45 Taking a Look Back at the Theory
50:40 Proofing the Fisher-Tipett-Gnedenko Theorem
52:21 Exercise: The Effect of the Block Size on the GEV Distribution
53:14 Solution: The Effect of the Block Size on the GEV Distribution
55:12 The Peaks Over Thresholds Method for Anomaly Detection
01:00:22 Exercise: Using the GPD for Anomaly Detection
01:02:03 Solution: Using the GPD for Anomaly Detection
01:08:55 More about Anomaly Detection
01:12:50 Optional Exercise: Anomaly Detection with EVT
01:13:49 Things That Were Omitted

The appliedAI Institute for Europe gGmbH is supported by the KI-Stiftung Heilbronn gGmbH.