Faculty Sponsor: Maryam Gooyabadi
Live Poster Session: Zoom Link
Aaron Foote
Aaron Foote is a rising senior from Chappaqua, New York. Prior to Wesleyan he attended Horace Greeley High School. Academically, Aaron studies computer science, data science, and applied math. He is excited to continue work on a computer science thesis under Professor Krizanc on computational complexity and puzzles come the fall, and is hoping to lead a student forum on sports analytics in the spring semester. Some non-academic interests include meditating, cooking, track and field, sports psychology, and reading. He serves as a QAC tutor and a course assistant for upper level computer science courses. After graduating from Wesleyan, Aaron plans to pursue a PhD in applied mathematics, operations research, or computer science.
Leave a comment if you have any questions, thoughts, or find the work interesting! I’d love to hear from you!
Abstract: Recent research in artificial intelligence, evolutionary game theory, and psychology have established Poker as a game of skill, creating an avenue for investigating dominant strategies in the Texas Hold’Em poker variant. The best of which established rational play as a dominant strategy in a pre-flop, heads-up game between rational and random agents. This project first presents a novel introduction to bluffing and proposes different dynamics for adaptive strategies in a population of rational, random, AAo bluffer, and T8o bluffer agents. Pairwise simulation of the four strategies within a death dynamic show bluffing to be a high variance strategy, and random play effective at winning hands but not money. Results indicate that strategies will struggle or thrive depending on the learning and population dynamics imposed as well as the composition of strategies present in the population.
The poster discusses learning dynamics and agent strategies that are not able to be entirely fleshed out on the poster. This PDF includes flow charts for each of the dynamics, flow charts for how each strategy determines their actions, tables explaining each dynamic/strategy in more detail, plus a definition of the agent class itself.
Aaron-Foote-Summer-Research-Poster-Final