Learning and Exploiting Recurrent Patterns in Neural Data

MISSING IMAGE

Material Information

Title:
Learning and Exploiting Recurrent Patterns in Neural Data
Physical Description:
1 online resource (196 p.)
Language:
english
Creator:
Brockmeier, Austin J
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Electrical and Computer Engineering
Committee Chair:
PRINCIPE,JOSE C
Committee Co-Chair:
LI,JIAN
Committee Members:
GUNDUZ,AYSEGUL
ACHE,BARRY W

Subjects

Subjects / Keywords:
algorithms -- decoding -- dependence -- electrophysiology -- metrics -- modeling -- neurons
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre:
Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Micro-electrode arrays implanted into the brain record the electrical potentials corresponding to the activity of neurons and neural populations. These recordings can be used to understand how a subject's brain represents different conditions such as external stimuli or movement intention. After learning this association, a subsequent condition can be decoded solely from the neural signals, enabling the brain to directly operate computers or machines, thereby creating a brain-machine interface. Brain-machine interfaces have the potential to improve as new technology enables concurrent recordings from an increasing number of signals throughout the brain. Neural signals are recorded at multiple scales: the action potentials, or spikes, from individual neurons, and the local field potentials corresponding to neural populations. On these diverse and high-dimensional signals, it is a challenge to pinpoint the indicators of different conditions. In addition, the neural responses have natural variability even for the same condition. Furthermore, it is likely that a portion, or even a majority, of the neural signals may pertain to other cognitive processes. To a naive decoder, this background activity appears as inexplicable noise. In this study, these challenges are addressed by proposing a set of methods that learn new representations of the neural data. These representations are adapted to both recurrent patterns in the neural signals and the decoding task. These methods include clustering and dimensionality reduction, which label or group reoccurring spatiotemporal patterns without supervision. Similarly, generative models are used to parsimoniously explain both the spatial and temporal patterns in neural potentials. In particular, models are explored that can account for variability in amplitude, waveform shape, and timing, and exploit spatial filters to separate different conditions. Finally, a new approach for optimizing the distance metrics for population activity is used to exploit information jointly represented across space and time and to highlight the most informative dimensions. Throughout the study, these tools were applied to neural recordings of both spike trains and local field potentials in different brain regions of animal models. The proposed approaches improve data visualization and decoding performance, aiding researchers in their quest to understand the brain from increasingly complex neural recordings.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Austin J Brockmeier.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: PRINCIPE,JOSE C.
Local:
Co-adviser: LI,JIAN.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2015-05-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2014
System ID:
UFE0046645:00001