G. J. Stephens

Deep Imputation for Skeleton Data (DISK) for Behavioral Science

DISK uses deep learning to impute missing skeleton data in animal behavior studies, improving reliability and enabling more accurate motion analysis.

A Markovian dynamics for C. elegans behavior across scales

This paper develops a Markov model of C. elegans foraging behavior that captures dynamics across scales, revealing new behavioral states and a trade-off between local and global search strategies.

Maximally predictive ensemble dynamics from data

This paper introduces a method to infer macroscopic dynamics from data by maximizing predictive information, applied to both physical systems and the movement of C. elegans.

WormPose: Image synthesis and convolutional networks for pose estimation in C. elegans

WormPose is an open-source tool for 2D pose estimation in C. elegans using synthetic data and convolutional networks, validated on both synthetic and real worms.

Flow-Mediated Olfactory Communication in Honey Bee Swarms

The paper studies how honey bees use directional communication through pheromone flow to aggregate efficiently during swarming.

Capturing the Continuous Complexity of Behavior in C. elegans

This paper develops a method to analyze C. elegans behavior by capturing its continuous state space, revealing both stereotyped and chaotic dynamics.

Modeling the ballistic-to-diffusive transition in nematode motility reveals variation in exploratory behavior across species

The paper models nematode motility transitions and finds species-specific variations in exploratory behavior linked to trade-offs between exploration and exploitation.

Pixel personality for dense object tracking in a 2D honeybee hive

The paper develops a method for tracking densely packed honeybees in 2D hives using pixel personality for object recognition, achieving 46% trajectory reconstruction in 5 min recordings.