Reading and writing neural computation : from silent speech to holographic optogenetics
Abstract/Contents
- Abstract
- This thesis presents a collaborative and interdisciplinary approach to understanding and manipulating neural computation across species. The work is divided into three main projects, each contributing to the overarching goal of improving human-computer interaction and deepening our understanding of neural dynamics. The first project explores Silent Speech Interfaces (SSIs) as a noninvasive alternative to brain-computer interfaces for soundless verbal communication. This work seeks to improve the performance of SSIs by using novel loss functions and large language models (LLMs) to significantly reduce word error rates in silent speech recognition on an open vocabulary. The development introduces Multimodal Orofacial Neural Audio (MONA) and LLM Integrated Scoring Adjustment (LISA), improving the state- of-the-art word error rate on silent speech from 28.8% to 12.2%. This demonstrates that SSIs are a viable alternative to automatic speech recognition and offers potential communication solutions for people with speech impairments. The second project focuses on translational research in SSIs at the intersection of chemical engineering and machine learning. We introduce a novel surface elec- tromyography (sEMG) sensor system that employs stretchable, sticky conductive polymers for high-fidelity adherence to the skin. In tandem with sensor development, this project also advances the application of high-speed state-space machine learning models for decoding sEMG on a 50-word vocabulary. This dual focus on advanced sensor technology and new machine learning methods opens new avenues for more reliable and human-centered wearable devices for biomedical applications. The third project shifts focus to the bioengineering of transgenic zebrafish to facilitate whole-brain optogenetics and holographic control of neural circuits. Through the creation of stable transgenic lines that express novel optogenetic actuators, this work enables the manipulation and observation of neural activity throughout the entire brain of a larval zebrafish. We also develop statistical and machine learning models to predict and denoise calcium dynamics in neural tissue. This project provides insight into a neural circuit involved in anxiety and depression, offering a new tool to dissect the functional organization of vertebrate brains. Together, these projects make advances in the fields of machine learning, bio- engineering, chemical engineering, and neuroscience in the pursuit of developing new modalities for human-computer interaction. From translational research in silent speech recognition for humans and basic research in neural circuit manipulation in zebrafish, this thesis contributes to the development of novel technologies for interfacing with the nervous system while drawing inspiration from the principles underlying neural computation.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2024; ©2024 |
Publication date | 2024; 2024 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Benster, Tyler Stephen |
---|---|
Degree supervisor | Druckmann, Shaul |
Thesis advisor | Druckmann, Shaul |
Thesis advisor | Baccus, Stephen A |
Thesis advisor | Bao, Zhenan |
Thesis advisor | Henderson, Jaimie (Jaimie M.) |
Thesis advisor | Linderman, Scott |
Degree committee member | Baccus, Stephen A |
Degree committee member | Bao, Zhenan |
Degree committee member | Henderson, Jaimie (Jaimie M.) |
Degree committee member | Linderman, Scott |
Associated with | Stanford University, School of Medicine |
Associated with | Stanford University, Neurosciences Program |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Tyler Benster. |
---|---|
Note | Submitted to the Neurosciences Program. |
Thesis | Thesis Ph.D. Stanford University 2024. |
Location | https://purl.stanford.edu/fs510zc9848 |
Access conditions
- Copyright
- © 2024 by Tyler Stephen Benster
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...