The volume of data we currently produce already challenges our ability extract and interprete information. From gravitational wave detectors to e-mail, automated collection of terascale data is now standard. Increasingly, progress in speed, resolution and information quality comes not from the capabilities of the hardware, but from the sophistication and useability of our algorithms. DDMA aproaches to the problem of intelligently extracting information from data include:

• Inverse Problems Inverse problems, or the task of obtaining model parameter(s) from observed data, are fundamental data analysis problems. DDMA works on the cutting edge in this area:
  • Compressed sensing
  • X-ray reconstruction
• Streaming Data We can expect that, while data processing time will keep pace with Moore’s law, data access time will not. Streaming algorithms that can process data when it is in transit (eg, as it is being collected or moved between memory locations) become increasingly efficient in this regime. One can take advantage of this efficiency, for example, by adaptively coupling data collection from scientific instruments and information extraction or by performing data reduction as data moves through a hierarchy of memory in a next-generation computer architecture.
• Measures and Metrics The problem of establishing a disciplined method for characterizing features of data sets and defining a method for comparative analysis of those features is fundamental to most lines of scientific inquiry. The DDMA team builds meaningful measures, metrics, and data transformations in some of the most challenging problems.
  • Theory
  • Methods
  • Applications
• Image Processing Image processing provides the foundation for much work in data analysis. DDMA has made contributions in this area from boundary extraction to de-blurring techniques.