Research Areas

Data Mining for Biological Applications
High performance, generic tools for data mining in bioinformatics; complex pattern mining, such as sets, sequences, trees, and graphs; algorithms for predicting protein folding pathways and misfoldings and matrix based algorithms for analyzing DNA microarray data. Funded by NSF, DOE.

Optimal Market Making Algorithms
Designing optimal market making algorithms for price discovery to make exchange traded markets more efficient. In particular studying the effect of different competition models (monopoly, duopoly, perfect competition) and different agent dynamics on the price dynamics.

Approximate and Randomized Matrix Algorithms
Developing sampling based randomized algorithms for revealing the structure huge (often low rank) matrices. In particular, identifying rank revealing representations and matrix decompositions (eg. CUR) and algorithms for applying these decompositions to mine information from large matrix arrays such as DNA microarray data. Funded by NSF early Career Award.

Social Network Analysis
Developing algorithms for modeling and analyzing hidden structure and information flow in large evolving social networks. In particular, identifying social groups and their dynamics, and how information flows through the social group structure. Funded by NSF, ONR and DHS (through its Center for Dynamic Data Analysis (DyDAn at DIMACS).

Astroinformatics
Identification of spatial structure in the Milky Way based observed luminosities and stellar density profiles. In particular, we develop efficient, highly parallelized (using Grid, WebComputing and Cluster approaches) probabilistic algorithms to extract parameterized geometric structures from geometric databases in an unsupervised manner.