Excited to be part of Project CETI working on machine learning and linguistics towards decoding sperm whale communication

Roboticists, biologists, linguists, and AI experts attempt to decode sperm whale communication. Very excited to be part of this team working on machine learning and linguistics.

A roadmap: arXiv paper

How does one approach a communication system of a species so different from us in which the unknown is not only what something means, but also how to test what something means?

Recent advances in speech and language technology (such as automated speech recognition, unsupervised translation, dialogue systems) will allow us to build machine learning models and train them on big data of sperm whale vocalizations in a similar way humans learn language — without supervision.

Using machine learning models and combining their outputs with decades of scientific work on human language, we will build descriptions of sperm whale communication and try to understand how their complex system of clicks corresponds to meaning.

Paper with Jacob Andreas, Gašper Beguš, Michael M. Bronstein, Roee Diamant, Denley Delaney, Shane Gero, Shafi Goldwasser, David F. Gruber, Sarah de Haas, Peter Malkin, Roger Payne, Giovanni Petri, Daniela Rus, Pratyusha Sharma, Dan Tchernov, Pernille Tønnesen, Antonio Torralba, Daniel Vogt, Robert J. Wood (alphabetically). Illustration: Alex Boersma