The course "Knowledge Discovery and Thread Detection", covers fundamental algorithmic techniques for knowledge discovery focusing on anomaly detection. In particular, we will study techniques to detect threats from stream-based data. First, we discuss about the basic concepts and tools used for anomaly detection (or outlier detection) in general and then we concentrate on algorithmic techniques for outlier mining from data streams. Then, we center our focus on topics related to CUREX, and more specifically on the application of state-of-the-art techniques for detecting unusual events.
The covered topics have as follows:
- outliers and anomaly detection
- introduction to data streams
- outlier mining over data streams
- threat detection in health data
- scalable anomaly detection (brief discussion of Apache Spark, Apache Flink)
- future research directions