For people living with disabilities that inhibit their ability to speak, feeling heard and understood can be a frustrating and difficult challenge. Yeganeh Marghi’s work with the Cognitive Systems Laboratory at Northeastern aims to make it less so through machine learning.
Marghi’s research has focused on the development of tools that can feed input received from a subject’s brain via EEG into a predictive algorithm, which can then translate their thoughts into words and actions faster and more accurately than ever before. She likens this process to playing Twenty Questions: “If I have twenty questions to guess what you are thinking of, if I am smart about it I should be able to guess correctly with fifteen or even ten questions. So I want to ask a minimum of questions, but I also want to be as accurate as possible in my guess. This is in essence an active query problem.”
By applying machine learning to the challenge of interpreting EEG signals, Marghi sees an opportunity to improve the quality of life of people relying on machines to move and speak by making it easier for those machines to determine what the user’s intent is. Beyond clinical applications, she also sees value in her research for people developing hands-free technology of all kinds, from manufacturing to communications and beyond.
Marghi credits Northeastern’s focus on collaboration as a great boon to her research; being able to work with professors and researchers in fields ranging from mechanical engineering to biology to physical therapy provided both different perspectives and new tools she was able to use for her own research. Plus, she believes her findings will be of value to them as well. “There are a lot of really smart people working in this field with backgrounds in biology, psychology, neuroscience. If they have someone with an engineering background or math background to develop this type of algorithm for them, many problems and challenges can be addressed.”
When she completes her doctorate in Spring 2019, Marghi will be moving to Seattle, to begin a research position with the Allen Institute.