- This event has passed.
Making models and mining mimics
Presented by: Chris Lawrence, Department of Ecology & Evolutionary Biology, Princeton University, Princeton, USA
Abstract: A common point of contact between biology and computer science uses machine learning (ML) and computer vision to classify species. For example, Inaturalist, Merlin, and Seek. Building upon previous work in this field, this talk will explore current work being done to expand ML algorithms to capture the biological mechanisms underlying Müllerian (where both species are chemically defended) mimicry in Heliconius butterflies. By modifying the training methods and inputs used in machine learning, computer vision is capable of asking biologically relevant questions. One approach, which modified the training method, trained classification on one Heliconius species before testing on its mimic, in order to approximate the training and learning process undertaken by avian predators. Another sought to account for the visual complexities of butterfly mimicry by adjusting the visual acuity of the images to better represent butterfly and bird vision. Finally, we quantified similarities and differences in the phenotypic spaces created by mimetic species. In the time remaining, I will talk about current work being done to identify traits and perform genotype-phenotype analysis.
Watch on ZoomWatch previously recorded webinars here
Thomas C. Emmel Seminar Series presents: Expanding Horizons in Lepidoptera Research
Continuing from Fall 2020, the McGuire Center is hosting this webinar series as an opportunity for both early career researchers as well as established leaders in the field to present their work. We hope that you will join us to hear about current advances in many diverse fields of Lepidoptera research.