“Last time you attended a crowded cocktail party, amid the talking and schmoozing, you could tune into a single conversation. To understand how we do this, we taught ferrets—whose auditory systems are remarkably like ours—to recognize a target word, and then put the ferrets in noisy, crowded situations and recorded their brain activity. Turns out the brain has a sort of neural volume knob that filters out background noise, letting them focus on a single voice. What we’ve learned could help improve hearing technologies.”
— Mounya Elhilali, Charles Renn Faculty Scholar and professor of electrical and computer engineering
Is more data always better?
“Using a method built on decades-old optimization theories, algorithms can now solve complex problems with only a few key pieces of information. For example, imagine optimizing FedEx package routes. While in the past, this required massive amounts of data, a new analysis now enables algorithms to estimate and fill in missing information while maintaining their general computational logic and accuracy. It’s like figuring out what a jigsaw puzzle depicts using only a few pieces. By avoiding analyzing unnecessary details, the method makes problem-solving more efficient and cost-effective.”
— Amitabh Basu, professor of applied mathematics and statistics
Can computer models simulate traumatic brain injury?
“Computer models enable us to do something extraordinary: map what happens to the brain under extreme impacts without putting anyone at risk. Using advanced imaging and machine learning, we can simulate events like car crashes and explosions, creating detailed models that show exactly how brain tissue deforms under severe stress. It’s like having a virtual crash test dummy for the brain, allowing us to understand potential injuries in ways we never could before.”
— Michael Shields, associate professor in the Department of Civil and Systems Engineering