CEBRA is a powerful machine learning tool designed for joint behavioral and neural analysis. It creates consistent and high-performance latent spaces from simultaneously recorded behavioral and neural data. CEBRA uses non-linear techniques to map behavioral actions to neural activity, uncovering hidden structures and patterns in time series data. This innovative approach allows researchers to probe neural representations during adaptive behaviors with unprecedented accuracy.
Major Highlights
- Produces neural latent embeddings for hypothesis testing and discovery-driven analysis
- Validated on calcium and electrophysiology datasets across species and tasks
- Supports single and multi-session datasets, with label-free functionality
- Offers rapid, high-accuracy decoding of natural movies from visual cortex
- Maps complex kinematic features in neuroscience research
- Creates consistent latent spaces across 2-photon and Neuropixels data
- Open-source implementation available on GitHub
- Improves decoding accuracy of behavioral variables over standard supervised learning
- Obtains embeddings robust to domain shifts
Use Cases
- Analyzing neural dynamics during adaptive behaviors
- Decoding behavioral variables from neural activity
- Mapping spatial representations in the hippocampus
- Uncovering kinematic features in motor cortex recordings
- Studying visual processing in the primary visual cortex
- Investigating neural correlates of behavior across species
- Developing brain-machine interfaces
- Conducting multi-animal and multi-session analyses
- Rapid adaptation to new, unseen neural data
- Exploring self-supervised learning in neuroscience
Leave a Reply