BugGuide.net is one of the busiest websites on Iowa State University's campus. This crowdsourced insect identification platform took in 17,548 images of insects and other arthropods in August 2017 alone. Specimens are classified by a skilled human volunteers. But might it be possible to replace the counting of tarsi or the agonizing over whether an antennomere is truly hirsute with the application of artificial intelligence? After all, Apple just announced a neural engine that fits in your phone and Google has been running machine learning networks since 2011. BugGuide has already displaced the use of traditional dichotomous keys. Can it take the next step and do identification without humans at all? Join John VanDyk, BugGuide maintainer and Levi Baber, director of research IT for LAS, to learn about what happened when ISU's high-performance computing cluster learned to identify insects.
Annotated presentation (35MB PDF)