Cebra

CEBRA is a novel machine-learning tool that enables the analysis of behavioral and neural data by compressing time series to reveal hidden structures within the data. It excels in processing data obtained simultaneously from behavioral and neural recordings and can accurately decode neural activity in the visual cortex to reconstruct viewed videos. CEBRA allows for the joint utilization of behavioral and neural data to uncover neural dynamics in a hypothesis-driven or discovery-driven manner and produces consistent and high-performance latent spaces. It is applicable to both calcium and electrophysiology datasets, across different sensory and motor tasks, and in a wide range of behaviors across species. This tool can be used with single or multi-session datasets for hypothesis testing or without labels. Additionally, CEBRA has the capability to map space, uncover complex kinematic features, and rapidly and accurately decode natural movies from the visual cortex.