Aditya Shrotri Aditya A. Shrotri
Ph.D. Student
Department of Computer Science
Rice University

3060 Duncan Hall
6100 Main St., MS 132,
Houston TX 77005-1892

E-mail: Aditya (d.o.t) Aniruddh (d.o.t.) Shrotri (a.t) rice (d.o.t) edu
Alternate 1: as128 (a.t) rice (d.o.t) edu
Alternate 2: aditya (d.o.t) a (d.o.t.) shrotri (a.t) gmail (d.o.t) com
CV     Code

About Me

I am a fourth year Ph.D. student at Rice University in the department of Computer Science advised by Prof. Moshe Y. Vardi. I am interested in designing provably efficient algorithms for problems arising from practical domains in Artifical Intelligence, with an emphasis on good empirical performance.... Techniques used widely in Formal Methods have remained under-explored in tackling problems in AI and Machine Learning. I hope to utilize my experience in working on both sides of the divide, for addressing modern AI challenges such as explainability and synthesis. Read more >>

Research Overview


The most interesting problems in AI are also the hardest, which has led to wide adoption of Approximation Algorithms that trade off accuracy for scalability. Recent trends show a bias towards achieving greater scalability using 'quick and dirty' techniques at the cost of worst-case guarantees. This is disastrous for safety-critical applications like ensuring the reliability of the electrical grid where the cost of failure is extremely high. In my current work on Approximate Model Counting, the goal is to design highly scalable approximation algorithms which do not compromise on accuracy.... For example, in our work on counting solutions to Boolean Formulas in Disjunctive Normal Form, we designed a hashing-based randomized approximation algorithm with PAC guarantees (a strong notion of accuracy) and linear running time. What set us apart from the then state-of-the-art was robust empirical performance across different benchmarks ensuring wider applicability than the other approaches. This abstract problem of DNF Counting has direct applications in Network Reliability among others, and having accurate and scalable tools for ensuring safety will be crucial in the future. Read more >>


Author names are sorted alphabetically


Before joining Rice, I worked for a year with Prof. S. Akshay on Model Checking of Markov chains as a Research Assistant in IIT Bombay, India. I did my Master's in Web and Data Mining from IIT Bombay advised by Prof. Soumen Chakrabarti. I was All-India Rank 1 among 150,000+ candidates in the Graduate Aptitude Test in Engineering (GATE) for Computer Science.