College of Engineering News • Iowa State University

Chinmay Hegde: Transforming big data with efficient algorithms

New ECpE assistant professor looks for better ways to handle data

Chinmay Hegde has always been curious about the power of math. He says that since he was young, he’s been interested in using math to “do cool stuff”. This interest has carried him through his education and now to Iowa State.

Chinmay Hegde’s education brought him all over. He completed his undergraduate at the Indian Institute of Technology Madras, studying electrical engineering. He then came to the U.S. for his M.S. and Ph.D. in electrical and computer engineering at Rice University in Texas.

“When I was a grad student, I started working on new mathematical techniques to do a better job at efficient data analysis,” he says. His postdoctoral work took him to MIT for three years before coming to Ames this fall to start in the Department of Electrical and Computer Engineering.

As an assistant professor, Hegde is teaching classes that include EE324: Signals and Systems II as well as completing research in the area of data analysis algorithms.

His work is at the intersection of a number of fields, such as machine learning, data science, algorithms and statistics. “The key idea here is that we’re being flooded by data from so many different sources,” he says. “How do you make sense of and get meaningful information from that data? And even more fundamentally, is it even worth acquiring all of this data in the first place?”

Answering that question allows researchers to determine the way they process or store the information. Hegde gives the example of determining the contents of a picture. “If you have a gigapixel image, and you simply want to determine if it has a tree in it or not, you don’t need to store all of the one billion pixels in it to determine that.”

Hegde’s goal is to create algorithms that are fast, scalable, and able to last in the quickly-changing world of technology. “We want algorithms that are ‘future-proof’ and will work ten to twenty years down the line. It seems to be a compelling problem, and that’s my motivation for looking into this.”