Resource Aware Ubiquitous Data Stream Mining Project

The world tends to mobility in its daily life activities. Ubiquitous computing has the same direction. Ubiquitous computing is moving the computational power to the area that surrounds the user. The main objective of ubiquitous computing is to free the user from time and space constraints. Ubiquitous Data Mining is one of the potential applications that motivate the need for ubiquitous computing. UDM is concerned with data analysis and delivery on mobile devices.

    Ubiquitous Data Mining (UDM) is the process of extracting hidden classifiers, clusters, frequent itemsets and association rules from data distributed among a number of mobile and stationary data sources. This research group is concerned with developing a resource-aware ubiquitous data mining system using different algorithmic and optimization techniques. The research progress and published research papers as well as the developed software will be available online.

     We have developed a resource-aware approach called Algorithm Output Granularity (AOG). AOG adapts to available resources according to input/output rates, memory and time constraints. Click here for a demonstration of the idea of AOG.

    Objectives

1)     Design and  Implementation of lightweight data mining, knowledge integration, and incremental learning algorithms for resource constrained devices.

2)      Design of a Resource-Aware Mobile Agent Framework for Ubiquitous Data Mining.

RA-UDM System Architecture