Distributed High-Dimensional Mining Tool for Bioscience Data Analysis

This SBIR effort focuses on developing an innovative data mining tool called ABMiner (Airborne Miner) to systematically analyze the huge amounts of bioscience data based on the synergy of distributed computation and the framework of prediction cubes that integrates clustering, classification and regression techniques comprehensively. Our research team includes Prof. Jiawei Han from UIUC and GDAIS. In Phase I effort, we have successfully completed the first version of prototype software. The current ABMiner enables users to define customized datasets, conduct a wide range of classification analysis using multiple approaches, conduct trade studies over those approaches, select classification models for knowledge representation, and ask what-if questions. The customization capability of ABMiner serves as the foundation for a flexible, extensible, and optimized data analysis tool for biomechanics. Our Phase II effort is the extension and refinement of the Phase I effort. The Phase II tasks include data integration and interface development, development of biomechanics scoring system, refinement of data mining capability using the framework of prediction cubes, refinement of user interfaces, development of web deployment solution, software implementation of ABMiner, validation with identified clients including AFRL and GDAIS, deployment of distributed information system, and technology transition and Phase III commercialization.