Anomaly Detection Algorithms and Techniques for Real-World Detection Systems

09:00 AM - 09:55 AM on July 17, 2016, Room CR5

Manojit Nandi

Audience level:

intermediate

Watch:

https://www.youtube.com/watch?v=CAvKQHHNmcY

Description

Finding outliers in a dataset is a challenging problem in which traditional analytical methods often perform poorly. As a result, researchers have developed special algorithms for detecting anomalies. In this talk, I will take about three different families of anomaly detection algorithms: Density-based methods, data streaming methods, and time series methods. I will cover both the mathematical and statistical theory behind these algorithms and provide code implementations. Afterwards, I will discuss useful tips I have learned while implementing threat detection systems in practice.

Abstract

Detecting anomalies is a problem that typically cannot be solved by traditional algorithms. In this talk, I will focus on three algorithms for detecting anomalies: Median Absolute Deviation, Local Outlier Factor, and Seasonal Hybrid Extreme Studentized Deviate. At the end of the talk, I will give advice on how to implement a real-world anomaly detection systems based on some of my experiences.