Viswanath Nagarajan
University of Michigan, MI·Computer Science
Many practical decisions have to be made before all the facts are in: a doctor orders tests without knowing the diagnosis, a retailer sets inventory without knowing demand, a robot plans a route through partly unknown space, and an inspector must decide which parts of a machine to check. In all of these, information arrives only as you act, so the question is not just what to decide but in what order to gather information.
- Fields
Computer Science
Mathematics & Statistics
Engineering
- Topics
approximation algorithms
stochastic optimization
sequential testing
- Grants
- 2 known grants, through Sep 2029
What they work on
- Design approximation algorithms with provable performance guarantees for sequential testing of systems made of interconnected components, including series-parallel structures, under complex cost structures.
- Characterize the tradeoff between batching tests together to lower cost and running them individually to obtain information faster, and evaluate the resulting strategies on publicly available datasets.
- Develop parallelizable algorithms for stochastic combinatorial optimization that commit decisions in a small number of sequential rounds, quantifying the loss in approximation quality relative to a fully sequential solution.
- Design algorithms for stochastic optimization when the input probability distribution is unknown, combining approximation algorithms with online learning, for core problems such as knapsack, set cover and influence maximization.
- Produce solution methods for stochastic dynamic programs with exponentially large state spaces, where exact computation is intractable.
More about the research
Work here designs algorithms for such problems and proves mathematical guarantees on how close their solutions come to the best possible, rather than relying only on empirical performance.
Why it matters
Better testing strategies could reduce the time and cost of inspection and quality assurance in manufacturing, infrastructure and healthcare technology, supporting more reliable systems and better-informed operational decisions. Algorithms that handle unknown or unreliable input distributions could improve planning under uncertainty in areas such as medical diagnosis, retail capacity planning, transportation and robotics, with provable guarantees rather than heuristics alone.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Sayan Bandyapadhyay
Portland State University, OR·Computer Science
Design polynomial-time approximation algorithms for balanced k-median and k-means clustering, where each cluster must fairly represent every protected group, aiming for constant-factor guarantees like those known for ordinary clustering.
Rajan Udwani
University of California-Berkeley, CA·Engineering·Early-career
Formulate new models of online resource allocation that capture the complex features, objectives and constraints of applications such as shared mobility, electric-vehicle battery swapping and online advertising.
Oliver Hinder
University of Pittsburgh, PA·Computer Science
Design stochastic optimization algorithms by first solving the noiseless problem and then extending them to noisy data with time-uniform concentration of measure results, instead of the standard online-to-batch conversion.
The school
University of Michigan: 124 labs on PI Pages, $2.1B a year on research, 963 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.