Streaming Data Analysis with Forgetting Factors
Title: Streaming Data Analysis with Forgetting Factors
Speaker: Prof. Niall Adams.Professor of Statistics, Department of Mathematics, Imperial College
Date: Thursday, 5th November 2015
Time: 4pm – 5pm
Location: Room E0.01, Science East
Abstract:
Automatic data collection is creating a need for improved streaming data analysis tools to support analysis and decision making in situ. The are two key challenges to address. First, there is the need for compute- and memory-constrained algorithms. The second relates to automatically handling temporal variation in evolving processes. The need for such automation is important because it is unrealistic to depend on human intervention. We are concerned with adapting standard statistical procedures to the streaming context by the incorporation of adaptively tuned exponential forgetting factors. This talk will review our work in this area, considering estimation, change-detection and classification.
Series:Statistics
Social Media Links