Two key themes characterize the bulk of NISLab ongoing research.
The first, data-starved inference for point processes, is focused on the development of statistically robust methods for analyzing discrete events, where the discrete events can range from photons hitting a detector in an imaging system to groups of people meeting in a social network. When the number of observed events is very small, accurately extracting knowledge from this data is a challenging task requiring the development of both new computational methods and novel theoretical analysis frameworks.
The second key theme of NISLab research, computational sensor analysis, involves using state-of-the-art computational tools to guide the design of novel sensors ranging from spectrometers to infrared imaging systems to hyperspectral imagers. The unconventional design of these sensors typically results in distorted, indirect measurements of the phenomenon of interest, and it is only by using novel computational tools which exploit the indirect nature of the data that we can extract meaningful information from the measurements.
The common thread which unites these research themes is that we can overcome the challenge of data scarcity by exploiting the sparsity inherent in many problems of interest. In these web pages, we describe these research themes in more detail, highlight specific key results, and describe interdisciplinary work with collaborators in astrophysics, biology, and photonics.