Towards the Amelioration of Classification Models for Evoked Potentials in Brain-Computer Interface by Chad Anthony Mello B.S., College of Santa Fe, Santa Fe 2009 M.S., University of Colorado Colorado Springs 2014 A dissertation submitted to the Graduate Faculty of the University of Colorado Colorado Springs in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Computer Science 2020 c 2020 Chad Anthony Mello All Rights Reserved ii This dissertation for the Doctor of Philosophy degree by Chad Anthony Mello has been approved for the Department of Computer Science by Terrance Boult, Chair Jugal Kalita Ethan Rudd Andrew White Yanyan Zhuang Date: December 16th, 2020 iii Mello, Chad Anthony (Ph.D., Engineering: Computer Science) Towards the Amelioration of Classification Models for Evoked Potentials in Brain-Computer Interface Dissertation directed by Professor Terrance E. Boult ABSTRACT Brain-Computer Interface technology has the potential to improve the lives of millions of people around the world. This study investigates how we may improve the performance of brain-computer interface for evoked potentials; we address some of the predominant chal- lenges that deter its widespread availability and application, demonstrating ways to augment system bootstrapping and performance with the adaptation of classifiers that seem better suited to generalizing across human electroencephalographic data. This dissertation intro- duces ways in which deep transfer learning, together with interpretability, may ameliorate the approach to deploying pre-trained brain-computer interface systems that generalize well across users and tasks. iv To the one who gave more than she had to give to this effort... Imelda, thank you for providing strength and encouragement on those days when I could find neither. Here’s looking to another twenty-seven years with you! Te adoro. Te necesito. Te Amo. ACKNOWLEDGMENTS I’d like to thank my PhD committee members for taking the time to participate in this effort. Some of you took extra time to learn about certain topics just so you could contribute here. Thank you for that! Each member on the committee was chosen because you brought something important to the table that my work ultimately benefited from in some way. Thank you for supporting and encouraging me through this journey. I’d like to take this opportunity to thank my advisor and committee chair, Dr. Ter- rance Boult, for his guidance and support throughout this process. Terry provided job opportunities, invaluable insight, wisdom, and expert knowledge in critical areas. Terry guided me towards putting all of this research into a coherent, cohesive body of work. Above all, Terry provided calm at times when I was about to commit Hara Kiri. Dr. Boult is undeniably a world-class researcher, and I am honored for having had the opportunity to work with him. I must also extend special acknowledgment and thanks to Ginger Boult who cleared the way for Terry to take on the advisory role for my work. She did this knowing she risked seeing less of her husband during that time period. Thank you, Ginger, for sharing some of your husband’s time with me. It means a lot to me, because I know it meant a lot to you. How could I not thank Ali Langfels for putting up with all my absent-minded mistakes and missed deadlines that undoubtedly made her life hell at times. Ali helped to keep me on track with accurate record-keeping, proctored my PhD examinations, and kept items that were important for graduating in front of my face the entire time. She made sure that I dotted all i’s and crossed all t’s. I would not be where I am here if she were not there. Thank you, Ali. vi A special thanks to Dr. Ethan Rudd who on a regular basis provided me with much insight through his experience and vast knowledge in machine learning, Python and beyond. He gave up many hours of personal time to allow me to bounce ideas off of him. His direct efforts resulted in much-welcomed funding that went towards this work, and it was a tremendous blessing. Ethan was instrumental beyond what I could express here. Thank you, Ethan. I’ve worked with Dr. Andrew White for many years now. It’s an honor to continue to work with him. Over the years he’s provided insight into how the brain works. Much of my knowledge in electroencephalography stems from the hours we spent on interesting and exciting research in epileptic brain function. Since then I’ve come to rely on Andy’s professional knowledge and understanding of the human brain. Thank you, Andy, for sharing your insight and for your continued interest and support of my work. I’d like to thank Dr.Yanyan Zhuang for agreeing to join my committee. From her feedback I’ve gained a lot of insight into how other people outside of BCI and machine learning might see this work. Yanyan passed along many pointers that helped me to prepare for my defense, and I honestly felt much more confident going into the defense after spending considerable time going over it with her beforehand. Thank you, Yanyan. Dr. Jugal Kalita provided valuable help over the years, and I’d like to thank him for preparing me for my defense by allowing me to give presentations on various topics related to my work to his graduate students. Presenting ideas and having those ideas challenged is the better part of what science and research is about. This hands-on approach helped me to become a better skeptic and debater. Jugal, thank you for the tough questions and for challenging my ideas. It all helped to bring the bigger picture into focus. vii I’d like to express a special thank you to Dr. Rory Lewis. I learned a lot by working with Rory throughout my grad school years. I never knew when Dr. Lewis might call me up in front of a crowd of people to speak, ask me to conduct a last-minute lecture for one of his classes, or when he might correct me while speaking in front of any audience. This kept me on my toes and forced me to become a better public speaker; a critical life skill for anyone looking to become anything in academia. I learned what it was like to author a book, to write and publish research papers, and to dream up wild ideas... all the things that make doing science fun. Some of the work we did together is covered in Chapter 3. Thanks for the wild ride, Rory. To the staff who make UCCS go, and those faculty members who dedicate themselves to teaching and mentoring the next generation of scientists and engineers, thank you for all your hard work, and thank you for extending the honor and privilege to me by allowing me to share in the experience through lecturing and teaching undergraduate courses at UCCS. I know your job can be tough at times, perhaps leaving you feeling discouraged on occasion. Take solace in the fact that you do make a difference in the lives of many people who are looking to create a meaningful and purposeful life on this earth. I’d like to thank the United States Air Force Academy CS faculty and staff for their encouragement and support since I started working there less than a year ago. They helped me to stay focused while encouraging me to put a lid on this work sooner rather than later. Specifically to Dr. Troy Weingart for dangling the carrot on a stick, prodding me towards an assistant professor position with the academy after graduation. Thank you! Two very good friends, Phillip Blanton and Terry Torres, provided me with nonstop energy and encouragement every step of the way. I want to thank them for demonstrating levels of positivity and enthusiasm such that I’ve never experienced before. That sort of viii tangible energy can drive people to accomplish extraordinary things. Gentlemen, thank you for sharing it with me. To Mom, Dad, Luke, and Amanda: Thank you for never balking at my crazy ideas, for all your patience and endless love. To my daughters, Tanya and Rachel, thank you for putting up with your Dad’s absentmindedness and crankiness over the last few years (as if I was ever anything but). My son and I missed out on a lot of time together in recent months. Thank you for understanding, David. I’m thankful that we still have a few more years left before you embark on your own adventures. You can bet I’ll be there to help you along. For all the times she had picked up the slack when I couldn’t be there, I want to thank my wife, Imelda. She never hounded, she never browbeat, and she always gave me the space and time I needed to work. She kept things together and maintained a sense of normalcy for the family when things were anything but normal. She was there every step of the way; she shared in the highs as well as the lows. Imelda, thank you for all your hard work, grace and poise throughout this journey. TABLE OF CONTENTS 1 Thesis, Motivation, Claims, and Contributions 1 1.1 Terminology.................................... 2 1.2 Motivation . 4 1.3 Claims....................................... 8 1.4 InvestigatoryItems................................ 9 1.5 Contributions................................... 10 2 Whither Goest thou, BCI? 11 2.1 A New Era of Technology: Fuel for Renewed Interest . 12 2.2 BCI: The ”Interface” in Brain-Computer Interface . 13 2.3 P300 BCI . 14 2.3.1 Trends in P300 BCI Application .