Florida State University
This project will help create data sets to train and test a machine learning algorithm that will detect anomalies in specimen data
How Weird is That? is training machines to recognize the language that researchers use to describe anomalies in the natural world (e.g., “in my experience it’s odd to see this species so far north”) and differentiate it from less alarming word uses (e.g., “the population consistently had an odd-number of segments”). Public participants are differentiating between these two cases and describing the abnormal feature as falling into one of several categories (e.g., the first example is a distributional anomaly). This enables earlier flagging of biological outliers to get the information to the right stakeholders faster. These oddities can mark the early stages of transformative change (e.g., an invasive species’ arrival), so faster information hand-offs are way better! The lead researcher on this project is an undergraduate student finishing her Honors Thesis in Computational Biology.