Tools

• Library: PyOD V2. Github. For multivariate data. A lib with multiple algo (45 so far).
  • Can use LLM to auto select model –> reduce manual tuning (suite non DS).
  • Work across multiple sectors: NLP (NLP-ADBench), TS (TODS), Graph (PyGOD), Distributed Systems (PyOD on Databricks) .
  • Benchmark: 45-page paper + AdBench, comparing 30 algo on 57 datasets.

Resources

• Anomaly Detection Learning Resources. Books + papers, courses, videos, datasets, os vs commercial, conferences.
• • Vision AI for anomaly detection: A quick overview