This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Large amounts of high-dimensional data not only create the need for analysis of the data and interpretation of results, but also the need for development of tools and methods that can handle such data. Many techniques are graphical in nature with ability to represent a small number of variables at a time. Application of information visualization using neural network techniques enhances knowledge extraction and is targeted towards complex high-dimensional data and provides for a very small, if any, loss of information. The approaches we work on as part of the project are based on a self-organizing map algorithm and are implemented in C/C++ programming languages using OpenMP (for shared memory) and/or MPI 2.0 (for distributed memory) libraries for high-performance computing. Placement of records in neural-network augmented classic visualizations is directed by all (or, if so desired, a subset of) dimensional values. This means that all dimensions contribute to the overall layout of records, not just a couple as this is the case in classic visualizations such as scatter plot, radviz, parallel coordinates, star plot, etc. While the focus of our efforts is on biomedical data sets, our algorithms have been successfully applied to a wide range of data sets from microarray to census data.
Showing the most recent 10 out of 179 publications
