TOWARDS EFFICIENT AND SCALABLE MACHINE LEARNING FOR FUTURE NEURAL INTERFACES
No Access Until
Permanent Link(s)
Collections
Other Titles
Author(s)
Abstract
Closed-loop approaches in systems neuroscience and therapeutic stimulation have the potential to revolutionize our understanding of the brain and develop novel neuromodulation therapies for restoring lost functions. Neural interfaceswith capabilities such as multi-channel neural recording, on-site signal processing, rapid symptom detection, and closed-loop stimulation are crucial for enabling these innovative treatments. However, current closed-loop neural interfaces are limited by their simplicity and lack of sufficient on-chip processing and intelligence. This dissertation focuses on the development of next-generation neural decoders for closed-loop neural interfaces, utilizing on-chip machine learning to detect and suppress symptoms of neurological disorders. These neural decoders offer high versatility, low power consumption, minimal on-chip area, and robustness against neural signal fluctuations. Chapter 2 explores migraine state classification using somatosensory evoked potentials, an emerging application for neural interfaces. In Chapter 3, we introduce a resource-efficient oblique tree model that enables low-power, memory-efficient classifiers for realtime neurological disease detection and motor decoding. Chapter 4 presents a novel Tree in Tree decision graph model with applicability beyond neural data, demonstrating success in general tabular prediction tasks. In Chapter 5, we propose an adaptive machine learning-based decoder to compensate for fluctuations in neural signals during test time. The dissertation concludes with a discussion of future research directions for on-chip neural decoders.
Journal / Series
Volume & Issue
Description
Sponsorship
Date Issued
Publisher
Keywords
Location
Effective Date
Expiration Date
Sector
Employer
Union
Union Local
NAICS
Number of Workers
Committee Chair
Committee Co-Chair
Committee Member
Shoaran, Mahsa