Dear all,
I’m writing to advertise our Multi-Agent Behavior Workshop and associated machine learning competition, aimed at developing better tools for the quantification of animal behavior:
The Multi-Agent Behavior Challenge on Unsupervised Learning is a competition to learn meaningful
representations of behavioral video datasets in a self-supervised way. Teams are given collections of video and pose data of interacting animals (mice, flies, or ant-beetle pairs), and asked to submit
an embedding that captures animals’ actions; embeddings are evaluated on a gauntlet of hidden tasks to judge their ability to capture human-defined behaviors. Round 2 has just launched, and runs through May 20th, with $6,000 in cash prizes, plus the first
50 teams to beat task baselines receive up to $400 in AWS credits.
Call for Papers: The Multi-Agent Behavior Workshop at
CVPR brings together researchers studying behavior of interacting agents, broadly defined (this year’s invited speakers spans topics in biology, neuroscience,
robotics, reinforcement learning, and autonomous driving.) We invite you to
submit your work to present at our workshop poster session, held on June 20th at CVPR in New Orleans.
Workshop organizers: Ann Kennedy (Northwestern), Jennifer J. Sun (Caltech), and Kristin Branson (Janelia)
Best regards,
Ann
Ann Kennedy
Assistant Professor
Department of Neuroscience
Northwestern University Feinberg School of Medicine
Chicago, IL 60611