Career Fair and Career Events

Title: “Decoding Spatial Attention in the Cocktail Party Problem with EEG and fNIRS”

Committee: Kamal Sen, PhD – BME (Advisor, Chair) Laura Lewis, PhD – BME Meryem Yuzel, PhD – BME Bobak Nazer, PhD – ECE

Abstract: The ability of a normal-hearing individual to effortlessly track the auditory stimulus of the target speaker shrouded by competing maskers is long recognized by scientists. This human phenomenon can be seen as the computational equivalent of three chronological events: determining the spatial location of the target stimulus, segregating the target stimulus from the irrelevant stimulus such as noise, and reconstructing the target stimulus from the mixture of stimuli. Existing algorithms to perform complex scene analysis are far from equal to human performance. Recently, a biologically plausible algorithm for source segregation based on physiological observations in the auditory system has been proposed. A critical input required by this algorithm is the spatial location of the target stimulus. This proposal will develop technology to decode and classify the spatial location of the target stimulus based on non-invasive recordings from the human brain and supplement the biologically plausible segregation algorithm. In particular, we will use electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS).