31 May 2022,
by Sarah Thomson, University of Stirling
At Stirling, I’m part of the Data Science and Intelligent Systems Research Group and my specialty is evolutionary computation (EC) – algorithmic approaches to optimisation which draw inspiration from natural evolution. As a postdoc, my main project has been using machine learning and evolutionary computation to build a system which personalises and optimises schedules for vocational rehabilitation. My PhD was on fitness landscapes, which are a mathematical tool to study how optimisation problems and optimisation algorithms interact.
I recently returned from a month-long research visit to Michigan State University (MSU) in the United States, which was facilitated by SICSA PECE funding. At MSU I visited, collaborated, and networked with Professor Wolfgang Banzhaf — a world leader in evolutionary computation — and Dr Kenneth Reid, a promising early-career researcher with a background in both EC and genetics.
During the research visit, I developed a prototype system which evolves different “variants” of exam-question code snippets using a field of EC called “genetic improvement” (GI). The GI can replace binary or unary operators, or delete or swap code lines. The underlying motivation for this is that hopefully in the future, a more developed version of this tool can produce a different variant of an exam-question code snippet for each student, to reduce the chances of cheating. Every variant evolved is different. At the moment, the prototype works for two exam-question programs. In the next stages of the project, we intend to add more questions, as well as extending ways that the code can be evolved. Additionally, we plan to publish an article on this work once it is sufficiently well-developed.
While at MSU, I presented a seminar to the Michigan portion of the globally-known BEACON (Center for the Study of Evolution in Action) consortium. This was well received, with many intriguing and inspired questions. In addition to this, I also networked at social events with MSU scholars, as well as a visiting scholar from Portugal.
Massive thanks to SICSA for facilitating this productive and highly enjoyable visit!