High Performance Computing Productivity Model Synthesis
International Journal of High Performance Computing Applications
A study of cross-validation and bootstrap for accuracy estimation and model selection
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 2
Security Concept and Implementation for a Cloud Based E-science Infrastructure
ARES '12 Proceedings of the 2012 Seventh International Conference on Availability, Reliability and Security
Hi-index | 0.00 |
Due to the recent explosion of research data based on novel scientific instruments and corresponding experiments, automatic features, in particular in data analysis, has become more essential than ever. In this paper we present a new Automatic Analysis Framework (AAF) that is able to increase the productivity of data analysis. The AAF can be used for classifications, predictions and clustering. It is built upon the workflow engine Taverna, which is widely used in different domains and there exists a large number of Taverna activities for various kinds of analytical methods. The AAF enables scientists to modify our predefined Taverna workflow and to extend it with other available activities. For the execution of the analytical methods, in particular for the computation of the results, we use our own cloud-based Code Execution Framework (CEF). It provides web services to execute problem solving environment code, such as MATLAB, Octave, and R scripts, in parallel in the cloud. This combination of the AAF and CEF enables scientists to easily conduct time-consuming calculations without the need to manually combine potential combinations of independent variables. It furthermore automatically evaluates all identified models and provides service for the scientists conducting the analysis. The framework has been tested and evaluated with real breath gas data.