Predicting long-term collective animal behavior with deep learning
Papaspyros, V.; Escobedo, R.; Alahi, A.; Theraulaz, G.; Sire, C.; Mondada, F.
Show abstract
Deciphering the social interactions that govern collective behavior in animal societies has greatly benefited from advancements in modern computing. Computational models diverge into two kinds of approaches: analytical models and machine learning models. This work introduces a deep learning model for social interactions in the fish species Hemigrammus rhodostomus, and compares its results to experiments and to the results of a state-of-the-art analytical model. To that end, we propose a systematic methodology to assess the faithfulness of a model, based on the introduction of a set of stringent observables. We demonstrate that machine learning models of social interactions can directly compete against their analytical counterparts. Moreover, this work demonstrates the need for consistent validation across different timescales and highlights which design aspects critically enables our deep learning approach to capture both short- and long-term dynamics. We also show that this approach is scalable to other fish species.
Matching journals
The top 2 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
- Collective Evolution Learning Model for Vision-Based Collective Motion with Collision Avoidance 96%
- Observation and simulation of chemically mediated searches in marine zooplankton 96%
- Prediction of Covid-19 spreading and optimal coordination of counter-measures: From microscopic to macroscopic models to Pareto fronts 95%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.