Undergraduate Research Program

Millicent Li

Millicent SmilingMajor: Computer Science
Mentor: Shwetak Patel, Computer Science; Noah Smith, Computer Science

Contact: limill01@uw.edu

Current research project: Multimodal Monitoring for Silent Speech Recognition; Debiasing Summarizations for Clinical Documents and Notes


Millicent is a senior studying computer science at the University of Washington. She was born in Wichita, Kansas, but attended both high school and now college in Washington. She is interested in the application of computer science to health, focusing on building tools and systems and creating algorithms to improve accessibility to healthcare. In her free time, she enjoys gaming, hanging out with friends, hiking, and traveling.

Translate your work so that we can all understand its importance
Speech is an incredibly useful modality for in-person conversation with others. However, speech can become limited for those with disabilities who are unable to speak, when privacy is necessary, or if ambient noise exists in the background. For my work, I’m creating a silent speech interface that can be used to create speech but does not require actual muscle movement, only fueled by intention. Current methods use a variety of signals including EEG and MEG. However, these signals are prone to large quantities of noise from movement and electrical interference. My work focuses on multimodal sensing using EEG and fNIRS, which has yet to be employed for work on silent speech interfaces. The end goal is to create a robust wearable that can be used in-the-wild for individuals who could use a tool such as this for speech. My other project focuses on developing methods to debias clinical notes and documents. Bias in the medical world can be caused by implicit racial/ethnical bias that can affect treatments for individuals. There has been some research on how we can take text from these documents and to screen them for bias, but my work concerns mostly the act of whether summaries from clinical notes and documents in themselves are biased and how can we create natural language models that remove these biases so that future treatments will not perpetuate these biases.


When, how, and why did you get involved in undergraduate research?
I actually first got involved in undergraduate research in high school (so really, high school research), by applying to a program in one of my labs (Ubiquitous Computing Lab) where I still do research. I was interested in figuring out what research was and what computer science research really entailed. I’ve gone through several mentors within the lab, but each time, have had an incredible experience getting to know everyone, from undergraduate to PhD students.


What advice would you give a student who is considering getting involved in undergraduate research?
Always reach out to professors and graduate students if you can! Everyone will be excited to talk to you about research since it’s such an amazing activity to take part in. Although I did get lucky to have a lab that I could come back to starting my undergraduate, I definitely wish I had spent more time reaching out earlier in my college years so I could explore more areas of research outside of my comfort zone. Luckily, I’ve found that focus on health is really my end goal, so it worked out for me. Also, never fear rejection from a lab. I’ve been rejected several times from research that I’ve really wanted, but it’s all part of the process. It might not feel good for a few days, but you’ll learn and be a stronger individual after rejection.