Adaptive Semi-Supervised Training of P300 ERP-BCI Speller System with Minimum Calibration Effort Restricted; Files & ToC

Abstract

A P300 ERP-based BCI speller is an assistive communication tool. It searches the elicited electroencephalogram (EEG) for target P300 event-related potential (ERP) responses among non-target alternatives. Existing methods require lengthy calibration, display static stimulus design without utilizing past EEG signals, and treat calibration and testing separately. Thus, we propose a unified framework incorporating calibration with minimum effort such that, given a small amount of labeled calibration data, we employ a semi-supervised adaptive EM-GMM algorithm to iteratively update a binary classifier. We evaluate our method using both simulated pseudo P300 signals and offline EEG recordings from real participants. Performance is assessed using character-level prediction accuracy, and results are compared against a supervised LDA method. Results of simulation and real data analysis indicate that the adaptive method outperforms the offline approach in our study. And semi-supervised learning can provide a practical and efficient alternative to fully supervised methods, particularly in contexts with limited labeled data.

Table of Contents

This table of contents is under embargo until 15 May 2027

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Biostatistics
Subfield / Discipline	Biostatistics - MPH & MSPH
Degree	M.S.P.H.
Submission	Master's Thesis
Language	English
Research Field	Health Sciences, Public Health Biology, Neuroscience Statistics
Keyword	Brain-Computer Interface (BCI), Calibration Less, Semi-supervised Learning, Adaptive EM-GMM Model
Committee Chair / Thesis Advisor	Tianwen Ma, Emory University
Committee Members	Max Lau, Emory University

Last modified

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	File download under embargo until 15 May 2027	2025-04-15 09:57:02 -0400	File download under embargo until 15 May 2027

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions