Ten Years of Living Machines Conferences: Transformers-Based Automated Topic Grouping.

The overwhelming growth in the number of scientific publications has its drawbacks: it is difficult to cope with it. Researchers need to constantly process the information contained in a very large and increasing number of articles to be up-to-date with their research community. This is especially difficult when interdisciplinary research requires knowledge of the state of the art in different fields. High interdisciplinarity is at the heart of the Living Machines conferences, which bring together scientists from different fields of biology, neuroscience, physics, robotics, artificial intelligence and other engineering fields. Here, using algorithmic AI methods, we analyze the publications of the ten "Living Machines" conferences to automatically discover clusters of research topics. We show that neural networks for natural language processing can find a set of thematic clusters that correspond to different research topics within this community.