Modeling and recognition of events from multidimensional data: doctoral symposium

Abstract

The recent rise in scale of sensors has led to the need for faster processing of events from multiple sensor data streams in a variety of real-world applications. We need an approach to model real-world entities and their interrelationships, and specify the process of moving from sensor data streams to event detection to event-based goal planning. Recent advances in analysis of temporal data, such as time series shapelets, provide methods for identifying these discriminative events for classification. In this dissertation, I make connections between event processing and time series data mining as part of a comprehensive event detection and representation framework.