Virginia de Sa
Professor of Cognitive Science, HDSI Chancellor’s Endowed Chair, and associate director of the Halicioglu Data Science Institute, UC San Diego.

Talk Title:
Recognizing, Creating, and Exploiting Multiple Views of EEG Data

Abstract:
EEG signals are known to be non-stationary and "noisy" which can reduce performance of brain-computer interfaces (BCIs).  In this talk I will show that some of this "noise" during BCI use reflects response to perceived task performance and can actually be used to improve performance in an EEG-based motor-imagery brain-computer interface (BCI).  We will discuss how this feedback response has improved robustness, relative to the motor imagery signal, when multiple steps are needed to complete an action.  We will also discuss using BCI techniques to predict whether people will remember pictures from EEG recorded prior to and during study presentation as well as during test presentation.  While this project was originally started to explore the idea of a passive BCI system to improve learning retention, the single-trial classification analysis also provides discriminative dimensionality reduction which reveals interesting electrophysiological distinctions between behavioral states.

Bio:
Virginia de Sa is a professor of Cognitive Science, HDSI Chancellor's Endowed Chair, and associate director of the Halicioglu Data Science Institute at UC San Diego. Her research goal is to better understand the neural basis of human perception and learning, both from a neural and computational point of view. She investigates what physiological recordings and the constraints and limitations of human performance tell us about how our brains perceive and learn.