As part of the upcoming Psychology Colloquium at Hobart and William Smith, Rajeev Raizada, assistant professor in the Department of Brain & Cognitive Sciences at University of Rochester, will deliver a talk about his brain imaging research on vision and language processing, as well as his innovative machine learning approaches to data analysis.
“Visual object perception: How machine learning and neuroscience can inform each other” will take place on Monday, April 11 at 4 p.m. in the Sanford Room of the Warren Hunting Smith Library.
“Professor Raizada is at the cutting edge of brain imaging research, having worked previously in top neuroscience research groups at Dartmouth College, Boston University and Cornell University.” says Assistant Professor of Psychology Daniel Graham.
During his talk, Raizada will present two studies testing probing neural representational structure. The first investigates how the brain’s structure changes when people learn to solve a 3D object-categorization task that has a linearly inseparable form. Researchers found that the brain learns to linearize the task, and that it does so in a way similar to how such problems are solved by algorithms in machine learning known as kernels.
In the second study, researchers examined how the brain represents different viewpoints of a 3D object. The retinal projections of these viewpoints are high-dimensional and variable, but they have a single underlying low-dimensional cause: they are all views of the same object. A continuous low-dimensional representation of this sort is known as a manifold. Computer vision systems use manifolds to represent different views of the same object. The researchers’ fMRI analyses indicate that the human brain may use the same computational trick.
“As neuroscientists, we often face challenges in explaining research findings to a general audience since the field draws on concepts from biology, mathematics, physics, computer science and other fields,” Graham says. “Yet our object of study – namely, how the brain does what it does – is of inherent interest to scientists and non-scientists alike. Professor Raizada is a very engaging and approachable speaker who has a talent for making advanced concepts understandable to a wide audience. This talk will be of interest to those in psychology, neuroscience, and the physical and mathematical sciences, as well as to those outside of the natural sciences.”
The event is sponsored by the Psychology Department, the Office of the Provost, and the Martin J. Kelly Endowed Discretionary Fund. The talk is free and open to the public. There will be refreshments preceding the talk at 3:45 p.m.