Markov Logic for Statistical Relational Learning

  • Parag Singla | Indian Institute of Technology

Classical machine learning makes the i.i.d. (independently and identically distributed) assumption on the data instances. Many real world problems have inherent relational structure where the i.i.d. assumption is no longer valid. Representing this relational structure explicitly becomes important for building accurate models of the data. Markov logic is a formalism to achieve the dual goal of representing the relational structure while handling uncertainty using well-founded statistical models. Markov logic models the underlying domain using weighted first-order formulas. These formulas are then compiled into a Markov network with features corresponding to the ground formulas and parameters corresponding to the associated weights. In this talk, I will present the theory behind Markov logic followed by inference and learning techniques. I will also briefly describe an application of Markov logic to the problem of online social network analysis.

Speaker Details

I recently joined as an Assistant Professor in the Department of Computer Science and Engineering at Indian Institute of Technology, Delhi. Earlier, I worked as a post-doctoral researcher with Professor Ray Mooney at the University of Texas, Austin. I finished my Phd with Professor Pedro Domingos at the University of Washington, Seattle in 2009.

I am interested in the area of Machine Learning. My research work lies around the problem of combining the power of logic and probability. I have worked on one such framework called Markov Logic. I am interested in the problem of efficient inference in such models and their application to the real life problems such as activity recognition in video. My other interests lie in the area of Social Network Analysis.

Watch Next