Data Mining and Machine Learning Series
Directed Acyclic Graph Model for Scalable Contextual Anomaly Event Stream Detection
25th January 2019, 10:00
Bakhtiar Amen
University of Liverpool
Abstract
The age of big digital data has emerged, the Internet of Things (IoT) and Internet of Everything (IoE) objects are creating high volumes of streams at high velocity rate. Thus, learning and detecting anomaly from continuous event stream poses a unique challenging task before it can be dismissed or disregarded. Specifically, retrieving contextual information from event stream is one of the primarily concern in many real-time applications including in network traffics, weather broadcast, stock exchange, healthcare sensors, engineering machines and online e-commerce. In this talk, I will demonstrates the anomaly detection concept for the big event stream scenario and present novel contextual detection method and algorithm to retrieve deeper information by using Direct Acyclic Graph (DAG) Model.
Maintained by Danushka Bollegala