Cebra: Learnable Latent Embeddings for Joint Behavioral and Neural Analysis

Cebra is a machine learning tool that utilizes non-linear techniques to create consistent and high-performance latent spaces from joint behavioral and neural data recorded simultaneously. It offers several key features:

🧠 Neural Latent Embeddings: Enables hypothesis testing and discovery-driven analysis.
🎯 Validated Accuracy: Proven efficacy on calcium and electrophysiology datasets, sensory and motor tasks, and simple or complex behaviors across species.
🔀 Multi-session and Label-free: Can be used with single or multi-session datasets and without labels.
📊 High-accuracy Decoding: Provides rapid decoding of natural movies from the visual cortex.
💻 Code Availability: Access the tool’s code on GitHub and read the pre-print on arxiv.org.

Use Cases

🔬 Analyze and Decode Behavioral and Neural Data: Uncover underlying neural representations through the analysis and decoding of behavioral and neural data.
🧠 Map and Uncover Complex Kinematic Features: Explore and understand complex kinematic features in neuroscience research.
🔁 Produce Consistent Latent Spaces: Generate consistent latent spaces across various data types and experiments.

Conclusion

Cebra is a valuable tool for neuroscientists seeking to analyze and decode behavioral and neural data. By leveraging non-linear techniques, it enables the exploration of underlying neural representations involved in adaptive behaviors. With its validated accuracy, multi-session and label-free capabilities, high-accuracy decoding, and code availability, Cebra empowers researchers to gain deeper insights into the complex workings of the brain.