I'm very grateful to be able to work with incredibly talented undergraduate students at Bridgewater College. Over the past 5.5 years I've had the privilege of working with 2 students on summer research projects, 3 students on their Honors projects, a group of 4 students in a research oriented course, and with another student while writing the online book. Below is a table that lists the students that I have had the honor of collaborating with as well as descriptions of the work that we've done.
During the Spring 2017 semeseter, the team extended the capabilities of YAGL to include directed graphs, weighted graphs, centrality measurements and created a new graphical user interface. Their work was presented at the 2017 A.S.P.I.R.E Poster Session. The new graphical user interface can be viewed here.
Trevor created a 3D simulation of an 8-bit computer using Unreal Engine 4 for his Honor's Project and Senior Seminar in the Fall 2016 and Spring 2017 semesters. The simulation is similar to the design discussed in the text "But How Do It Know?" by J. Clark Scott and was inpired by the large scale Megaprocessor built by James Newman at Cambridge. As Trevor writes in his proposal, "the goal is to create a simulation that will help students learn and understand how a computer functions at a low level." His work can be viewed in the videos he produced and in his 2017 A.S.P.I.R.E poster.
Dylan Tokotch's Honor project, conducted during the Fall of 2015, was originally titled “Normalizing Carotid Pulses Using Dynamic Time Warping (DTW) And It’s Use in Biometric Identification”. Shortly after the research began, however, he found a paper which suggested that DTW was not helpful in thoracic biometric identification, so he changed course and instead of implementing DTW in the thoracic biometric software, Dylan implemented the heart beat detection algorithm written by Bridgewater College Mathematics Professor Dr. Verne Leininger and student Jose Corona for their 2015 Research Experience @ Bridgewater College grant. Dylan’s work resulted in a recognition rate of 69% (+4). His work was documented in his Honors paper and in the presentation he gave to his committee members. We are currently writing a paper describing the heart beat detection algorithm and our results. We hope to submit it for publication in 2019.
In this work Troy performed an IRB approved data collection to collect carotid pulse signals from multiple subjects. Each of 15 subjects were tested in two sessions, with sessions approximately two weeks apart. During each session he collected pulse signals before and after exercise. Troy also worked on the TIS software and increased the carotid pulse recognition rate to 65% (+2). He presented his work and results at the 2016 A.S.P.I.R.E. poster session and the data collected can be found on GitHub. This research was funded by a 2015 Research Experience @ Bridgewater College grant.
In this project, Tayseer extended my post-doctoral research by modifying the biometric software that I wrote to identify individuals using thoracic signals. The software is named Thoracic Identification System or TIS. Among the many modifications to the software that Tayseer made were a rewriting of the beat detection algorithm and the beat selection algorithm. Her modifications increased the carotid pulse identification rate from 23% to 63% (+40). She presented her results at the 2015 A.S.P.I.R.E. poster session and a report was sent to the Intelligence Community agency that sponsored the original study.
The goals of this research project were to determine biometric performance measurements for the acoustic, electrical and pulse signals produced by the heart, referred to as thoracic signals, and to identify if a relationship exists between the electrical and pulse signals. Our results can be found in our final report and the Thoracic Identification System was released as open source.
The first refutationally complete inference systems for first-order logic, called instance-based systems, were based on Herbrand’s theorem which implies that first-order logic satisfiability can be reduced to propositional logic satisfiability (SAT). Out of this line of research came the landmark SAT solving DPLL algorithm. Soon after DPLL made its debut, Robinson introduced the simple combinatorial resolution rule which detracted interest in instance-based systems. Recently, with the increase in computational power of the personal computer, there has been renewed interest in systems for first-order logic theorem proving that utilize SAT solvers. In my dissertation we present three novel solutions for the first-order logic validity problem that utilize SAT.