Cebra

CEBRA is an innovative machine-learning tool that connects behavioral actions to neural activity, empowering researchers in neuroscience.

CEBRA, which stands for Learnable Latent Embeddings for Joint Behavioural and Neural Analysis, is a cutting-edge method designed specifically for mapping how our actions relate to brain activity—one of the major goals in neuroscience. With the growing ability to collect extensive data on neural and behavioral activities, CEBRA directly addresses the rising demand for tools that can model these complex dynamics effectively.

What sets CEBRA apart is its versatility. It can utilize both behavioral and neural data in two main ways: it can be driven by a specific hypothesis or it can help discover new insights without preconceived notions. This flexibility allows researchers to create accurate and reliable latent spaces, shedding light on the connections between behavior and the brain.

This tool is also incredibly adaptable, working seamlessly with datasets from single and multiple sessions. Whether you're testing a hypothesis or exploring data without specific labels, CEBRA can handle it. Additionally, it's compatible with different types of neural data—both calcium imaging and electrophysiology—making it suitable for various tasks, whether they involve sensory input, motor functions, or even complex behaviors across different species.

One of CEBRA's standout features is its ability to map spaces, reveal intricate kinematic patterns, and rapidly and accurately decode visuals from the brain's visual cortex. This capability significantly enhances our understanding of how neural dynamics relate to behavior. For instance, it excels at decoding the activity within the mouse brain’s visual cortex to reconstruct videos that the animal has seen, showcasing its potential to contribute meaningfully to both neuroscience and behavioral research.