Modeling animal movement to decipher behavior using machine learning techniques

This research spans across neuroscience, biology, physics and computational sciences to address novel questions in developmental biology, cerebellar function, and behavior. This research will ask the following questions: 1) How does social behavior differ between a classic three-chamber task and a group social interaction? 2) What unique aspects of social interactions can be used as a predictor of future social behavior? 3) Can we dissociate typical rodent behavior from those with perturbations? Students will learn how to track and analyze recorded animal behavior using machine learning tools.

The educational goals of this program are to teach students about animal behavior and modern neuroscience tools, to provide them with critical experience for their future careers. Students will have the option of attending lab meetings, where primary literature will be read, critically analyzed, and the statistical analysis will be explained. Each student will meet one-on-one with their graduate mentor biweekly and each team will meet with Dr. Verpeut monthly to discuss research goals, data, and analysis.

At the end of the program, students will fulfill the following learning objectives: 1) Become proficient in the Python coding language, 2) Learn how to utilize SLEAP to track animal joint movement and analyze those movements using clustering techniques in keypoint-Moseq, 3) Understand the rodent social dynamics, 4) Be able to read and critically analyze the primary literature, 5) Have conceptual understanding of statistical analyses of behavior using machine learning and mathematical clustering models. This work will require a time commitment from each student of 10 hours per week, including meetings. Students will sign up to receive 3 research course credits for participation (NEU 499 or Barrett Honors credits).