Current Research

 

I am a Ph.D. student at the University of Rochester working on topics closely related to Markov logic, and probabilistic graphical models under the supervison of Prof. Henry Kautz.

The topics in a nutshell:


Hard constraints given as first order logic formulas in Markov logic can force truth values on ground literals. With these new set of ground literals we may not have to instantiate the soft constraints in the network for certain groundings saving a lot both on the space and time complexity of the inference algorithm. (Project with Prof. Henry Kautz and Prof. Parag Singla, IIT, Delhi.)

Incorporating the knowledge of a domain expert into Markov logic networks is not straightforward, because the natural parameterization, i.e., the weights of formulas in the knowledge base are hard to relate to the (conditional) probabilities of formulas. We show that with the proper reparameterization the expert's (possibly inconsistent) knowledge can be combined with the data within the framework of Bayesian statistics. (Joint work with Dr. Shalini Ghosh, SRI International and Prof. Henry Kautz.)

The marginal probabilities of ground atoms when Markov logic networks are unrolled into conditional or Markov random fields depend on the size of the domain. We restrict our research to sequential (temporal) data. We propose a different discriminative graphical model we can unroll our Markov logic theory into which does not have certain undesired theoretical weaknesses of MRFs/CRFs, and allows for implementation of efficient particle filtering. (Work with Prof. Henry Kautz.)

For Bayesian networks we are developing an algebra with a special multiplication operator. We can represent the joint probabilities of network nodes in the Bayesian network as algebraic expression. Our short term goal is to create efficient (exact and approximate) inference algorithms using this new algebra. (Work with Prof. Len Schubert.)