Wikimedia Commons

Amin Karbasi, assistant professor of electrical engineering, computer science and statistics and data science, and Mehraveh Salehi GRD ’19 came in second place in the Bell Labs Prize for their proposal of a data modeling method for generating personalized maps of patients’ brains.

The Bells Lab Prize is an annual competition held by Nokia Bell Labs to find innovations in science, technology, engineering and mathematics that tackle modern challenges, such as those related to the telecommunications industry. Karbasi and Salehi were notified of their award on Nov. 9, winning a cash prize of $50,000. In 2021, more than a hundred applicants participated in the competition.

“Everyone really had an awesome story, a very impactful research, a very impactful proposal,” Salehi said. “When we got the second place, I was pretty happy, pretty excited. … Among all these awesome works that I saw, I’m so glad to be part of innovators.”

According to Salehi, the idea for the project came in 2017. Karbasi noted that they strived to determine whether artificial intelligence could help give better insight towards understanding human brains. Karbasi and the research team wanted to explore whether they could “connect natural intelligence and an artificial intelligence.”

Previously, researchers were constructing “one atlas to … explain everyone,” according to Salehi. 

Karbasi noted that they used fMRI data, which collects information about oxygen levels in different parts of the brain to conduct their analysis. He mentioned that “when people perform different tasks, the oxygen levels in different parts of the brain” change, and they explored whether these changes were significant enough to help “read the brain.” In their research, they discovered that individuals could be identified with 99 percent accuracy through looking at their brain images. This led them to examine whether they could construct maps of the brain of an individual in order to predict what the person was doing during the fMRI. According to the results, the researchers were able to do so about 97 percent of the time.

Karbasi noted that the collaboration between Salehi; Dustin Scheinost, associate professor of radiology and biomedical imaging; and Todd Constable, professor of radiology and biomedical imaging, neurosurgery and biomedical engineering, was key to the success of the project. He emphasized that Salehi played an important role in the group effort as she “was basically the link that made the whole thing possible.”

“She’s probably one of the most, most amazing people I’ve ever seen,” said Karbasi. “She was one of those people who wanted to work on high risk, high reward projects. … She was a person that wanted to take up the challenge. She started working very hard.”

Salehi recalled the experience of working with Karbasi as a “very fun journey.” At one point, she struggled with figuring out how to use the brain’s biological information to construct maps. In solving this problem, Salehi noted that she enjoyed being able to speak with her advisors, including Karbasi, about how to connect the two domains of algorithms and brains. 

Other professors at Yale who have worked with Salehi echoed similar sentiments about her achievements. 

“Salehi took my linear systems course while she was at Yale — and did well,” A. Stephen Morse, Dudley professor of electrical engineering, wrote to the News. “She was a very good student!”

In terms of the prize money, Karbasi noted that they had not yet decided what to do with it. According to Karbasi, he views the prize as the “marshmallow test” where they need to decide whether “to have fun right away or to invest it and have fun later.”

Dionysis Kalogerias, assistant professor of electrical engineering, has collaborated with Karbasi on research projects and emphasizes his strength in understanding a variety of different subjects.

“He has worked in quite a few different topics, and his understanding of those topics, like broadly, is very nice,” Kalogerias said. “This is an important quality … in an academic. You don’t want to be very, very focused on something. What you want to do is to understand the potential of taking an idea, putting it in another context or combining things.”

Kalogerias’s observation of Karbasi’s qualities coincided with advice Karbasi has offered students looking to conduct research or to decide on a field of study. Karbasi emphasized the importance for students to “[be] broad” by trying “to understand as much as possible about different fields.”

Karbasi recognized that his advice may go against what many disciplines do, which is to encourage students to be “extremely specialized.” According to him, the prevailing advice is “a mistake,” and students should always “be ready” to apply their understanding of multiple fields in their education.

“The amazing thing at Yale that I really enjoyed is that … students have a broad interest,” Karbasi said. “They always amaze me. … They come with these crazy ideas. You just need one of these ideas to work … Yale undergrads, they’re super good at having all sorts of these ideas.”

Salehi advised students to reach out for support when they need help. According to Salehi, “everyone you know,” such as peers, are going through the same experience. She suggested that when students are confused, it is important to ask questions so the people around them can help. 

In addition, Salehi emphasizes that students should pursue what they are really interested in. She recalled that when she started her doctorate, she was working on something that she “basically had no interest” in. Then, she searched her lab for a project that she would really enjoy, and she looked at fields like biomedical engineering. 

“Really go for something that you enjoy doing because that is really important,” Salehi said. “Because then you would enjoy your time and also you would be the best of yourself, so the chances of success develop higher.”

The Bell Labs Prize awards cash prizes of $100,000 to the first-place winner, $50,000 to the second-place winner and $25,000 to the third-place winner.