by Arun Kejariwal | Publisher: Infinite Skills | Release Date: June 2017 | ISBN: 9781491983676

Anomaly detection plays a key role in today's world of data-driven decision making. This stems from the outsized role anomalies can play in potentially skewing the analysis of data and the subsequent decision making process. This course is an overview of anomaly detection's history, applications, and state-of-the-art techniques.Taught by anomaly detection expert Arun Kejariwal, the course provides those new to anomaly detection with the understanding necessary to choose the anomaly detection techniques most suited to their own application. While not required, a basic understanding of statistics, R, and Python will be helpful to get the most out of the class. Survey the history of anomaly detection in astronomy, statistics, and manufacturing Gain a core understanding of the most important anomaly detection techniques available today Explore the landscape of applications where anomaly detection is routinely used Develop an awareness of the underlying assumptions and challenges of anomaly detection Learn how to mitigate the influence of anomalies during data-driven decision making processesArun Kejariwal is a Statistical Learning Principal at Palo Alto based Machine Zone, where he leads R&D teams working on novel techniques for fraud detection and real-time anomaly detection. He developed many open sourced techniques for anomaly detection and breakout detection while working for Twitter; he speaks frequently at the Velocity and Strata Data conferences, and he's the co-author of the O'Reilly title "The Art of Capacity Planning: Scaling Web Resources."

1. Introduction
   • Welcome To The Course 00:03:19
   • About The Author 00:01:52
2. Historical Overview
   • Overview Of Anomaly Detection 00:10:20
   • Early Works - Astronomy 00:04:57
   • Early Works - Statistics 00:07:45
   • Early Works - Manufacturing 00:07:57
   • Robust Statistics 00:09:57
3. Applications
   • Applications Overview 00:01:12
   • Example Applications Part 1 00:11:37
   • Example Applications Part 2 00:16:25
4. Techniques
   • Overview Of Anomaly Detection Techniques 00:05:59
   • Supervised vs. Unsupervised Learning 00:15:38
   • Time Domain vs. Frequency Domain Part 1 00:13:00
   • Time Domain vs. Frequency Domain Part 2 00:05:31
   • Parametric vs. Non-Parametric 00:03:17
   • Probability Theory 00:01:21
   • Information Theoretic 00:02:36
   • Neural Networks 00:04:16
   • Graph Anomalies 00:06:07
5. Going Forward
   • Industry Initiatives 00:04:07
   • Open Problems 00:06:41
   • Multi-Dimensional Anomaly Detection 00:07:00
   • Interpretability 00:04:17
   • Streaming Vs. Real-Time Data 00:01:32
   • Next Steps 00:02:08
6. Conclusion
   • Wrap Up And Thank You 00:05:32