HOME ABOUT RESEARCH CONTACT
ZOE STEINE-HANSON

ABOUT ME

I love brains and computers

I am a NSF Graduate Research Fellow and PhD student at the University of Washington in the department of Computer Science and Engineering, currently co-advised by Dr. Bingni Brunton and Dr. Rajesh Rao. After completing my Honors Bachelors of Science in Computer Science at Oregon State University, I came to Seattle to learn more about the fascinating intersection between Neuroscience and Machine Learning. I first gained interest in the brain, and intelligence in general, during an undergraduate research experience at the University of Washington with Dr. Andrea Stocco, focusing on Cognitive Neuroscience. Currently, my research aims to improve generalizability of Brain Computer Interfaces (BCIs) to new patients by leveraging transfer learning techniques from machine learning on naturalistic electrocorticography (ECoG) data.

While at OSU I was involved with research in Human Computer Interaction with Dr. Margaret Burnett, specifically investigating gender biases in user interfaces. My Honors undergraduate thesis investigated "best practices" and design guidelines for gender inclusive software.

You can find a copy of my CV here

My Skills:

Data Analysis (python)

80%

Deep Learning (tensorflow)

75%

Writing

85%
RESEARCH

MY WORK

Here are some of my projects, in chronological order from most to least recent.


Spectrograms for different movement directions show that upward movements have best signal.

Transfer Learning with ECoG: My first project as a graduate student looked to improve brain decoders for electrocorticography (ECoG) signals through the use of transfer learning. We found that models pretrained on a pool of subjects and then fine-tuned to the target subject had comparable performance to models trained on all the target subject's data. Slides on our findings can be seen here, and links to the paper and github repo

Fine-tuning HTNet improves performance, even when few training events are available.

Deep -Um- Discourse: I was the team lead for my senior capstone project which developed a machine learning tool to detect filler words in speech. Using Mozilla Deepspeech, we were able to adapt the network to a small dataset we collected containing filler words in speech.

Common Model of Cognition: I spent a summer with Andrea Stocco at the University of Washington investigating the efficacy of the Common Model of Cognition (CMC) for human brain data. Using publicly available fMRI data from the Human Connectome Project, I was able to determine that the CMC most accurately explains brain activity over other possible models. See my poster from the summer and my 5 minute talk for more details.

GenderMag: I investigated possible design solutions for two cognitive styles (information processing and learning), which was the focus of my undergrad honors thesis. The video here shows these two styles in action, and how they influence the way users interact with software. Visit the GenderMag website for more details.

CONTACT

HAVE QUESTIONS?

University of Washington, Seattle, US
Email: zsteineh@uw.edu