Decoding EEG and MEG Signals
I'm very pleased to welcome Sahar Akram, who graduated and performed a postdoc at the University of Maryland. In concert with Behtash Babadi, she has developed and will talk about two algorithms for deciding which signal a subject is attending, as well as new dynamic decoding algorithm. In past years we've talked about simple linear models of the audio to EEG system. She extends these models by making the system respond to changing conditions, as well as putting the attention decision into a proper probabilistic framework. Both directions are important for those of us thinking about understanding the brain from EEG, MEG and eCog signals
Who: Sahar Akram (Univ of Maryland, and now Starkey Research)
When: Friday November 18th at 10:30AM
What: Decoding EEG and MEG signals with Probabilistic Models
Where: CCRMA Seminar Room
Why: We need better ways to understand EEG and MEG signals
Bring your favorite brain to the Stanford Hearing Seminar, and we'll talk about how we can read your mind---or at least talk a bit more intelligently about the information present in EEG and MEG signals. This is an important topic for those interested in EEG/MEG signals and/or probabilistic signal processing.
Decoding EEG and MEG signals with Probabilistic Models
Dr. Sahar Akram (Univ. of Maryland, and now Starkey Research)
Abstract: Decoding the dynamics of brain activity underlying conscious behavior is one of the key questions in systems neuroscience. Sensory neurons, such as those in the auditory system, can undergo rapid and task-dependent changes in their response characteristics during attentive behavior, and thereby result in functional changes in the system over time. In order to quantify human's conscious experience, neuroimaging techniques such as electroencephalography (EEG) and magnetoencephalography (MEG) are widely used to record the neural activity from the brain with millisecond temporal resolution. Therefore, a dynamic decoding framework on par with the sampling resolution of EEG/MEG is crucial in order to better understand the neural correlates underlying sophisticated cognitive functions such as attention. I will talk about two recent attempts on real-time decoding of brain neural activity during a competing auditory attention task, using Bayesian hierarchical modeling and adaptive signal processing.
Dr. Sahar Akram received her Ph.D. and M.Sc. degrees in Electrical and Computer Engineering from University of Maryland in 2011 and 2015, respectively, and the B.Sc. degree in Electrical Engineering from Sharif University of Technology, Iran in 2009. From January 2015 to August 2016, she was a postdoctoral fellow at Electrical and Computer Engineering department at University of Maryland .
Sahar's M.Sc. and Ph.D. thesis focused on neural and computational approaches to auditory scene analysis, with applications to auditory brain computer interfaces. Her main research interests are auditory neuroscience and statistical auditory signal processing.