Cebra

About Cebra

CEBRA (Consistent Embeddings of high-dimensional Recordings using Auxiliary variables) is a cutting-edge self-supervised learning algorithm designed to analyze complex neural and behavioral datasets.

It effectively compresses time-series data from recordings such as 2-photon and Neuropixels, uncovering hidden structures and variability that are often challenging to detect.

CEBRA excels in decoding neural activity related to behavior, such as reconstructing videos from visual cortex data, decoding trajectories in sensorimotor areas, and mapping spatial navigation.

Its versatile application spans across species, including mice, rats, and primates, using data from different brain regions like the visual cortex and hippocampus.

The tool produces interpretable, consistent latent spaces that can be used for hypothesis testing, decoding, and understanding neural dynamics during behavior.

Its ability to handle multi-session data, produce rapid high-accuracy decoding, and generate attribution maps makes it a powerful resource for neuroscientists and researchers interested in neural representations, brain-computer interfaces, and time-series analysis.

Developed and patented by EPFL, CEBRA is openly available for academic use, with ongoing research enhancing its capabilities..

Smart Features

• Self-supervised learning for high-dimensional neural and behavioral data
• Reveals hidden structures and variability in time-series data
• Accurate decoding of neural activity related to behavior and stimuli
• Produces consistent, interpretable latent embeddings
• Applicable across multiple species, brain regions, and recording modalities
• Supports multi-session data analysis for hypothesis testing
• Open-source implementation with ongoing updates and enhancements
• Patent-pending technology for advanced dimensionality reduction

Use Cases & Applications

• Decoding visual stimuli from mouse visual cortex recordings
• Mapping neural trajectories during motor tasks in primates
• Analyzing spatial navigation and position decoding in rodents
• Uncovering neural correlates of behavior across sensory and motor regions
• Multi-session neural data analysis for hypothesis testing
• Generating attribution maps for time-series data
• Brain-computer interface development
• Understanding neural representations in complex behaviors

Who is it for?

• Neuroscientists studying neural dynamics and behavior
• Machine learning researchers focused on time-series data
• Brain-computer interface developers
• Computational neuroscientists
• Behavioral scientists analyzing neural correlates
• Data scientists working with multi-modal neural datasets

Business Opportunities in Cebra

Leveraging CEBRA's advanced neural data analysis capabilities opens opportunities for consulting services in neuroscience research, developing neural decoding applications, and creating educational tools for brain-computer interface startups.

Companies can offer custom analysis pipelines for behavioral and neural data, enhance neurotechnology products with high-precision decoding, or provide training workshops on cutting-edge machine learning techniques for neural data.

Additionally, integrating CEBRA into neurofeedback or cognitive enhancement platforms can create innovative health tech products, fostering new revenue streams in the growing neuroscience and AI markets..