U-M Industrial and Operations Engineering assistant professor, Viswanath Nagarajan, has received funding from the National Science Foundation (NSF) to explore models and algorithms for stochastic covering problems in the presence of noisy outcomes.
“A common task in medical diagnosis, manufacturing and emergency-response applications is to perform sequential tests to identify the presence and/or locations of defects as quickly as possible,” said Nagarajan. “Classical models for these problems assume that random outcomes are observed without any noise. In practice, however, many of these outcomes are unknown or noisy.”
“Noisy” outcomes are test-outcomes that do not exactly match their predicted outcomes. A major limitation of existing models is that they assume that test-outcomes will not be noisy. Since noisy outcomes are common in practice, existing models produce policies that perform poorly when applied.
By accounting for noisy outcomes, Nagarajan’s research has the potential to improve both the accuracy and efficiency of test algorithms regardless of the application.
“It feels great to receive NSF support to work on a new class of stochastic optimization problems. This will allow me and my graduate students to design and test algorithms for some fundamental stochastic models in the presence of noisy/unknown outcomes, an area where there is very little existing research,” he said.