Cebra is a machine learning tool that uses non-linear techniques to create consistent and high-performance latent spaces from joint behavioural and neural data recorded simultaneously.

Key Features:

• Neural Latent Embeddings: Use for hypothesis testing and discovery-driven analysis.
• Validated Accuracy: Efficacy proven on calcium and electrophysiology datasets, sensory and motor tasks, and simple or complex behaviours 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 visual cortex.
• Code Availability: Access the tool’s code on GitHub and read the pre-print on arxiv.org.

Use Cases:

• Analyze and decode behavioural and neural data to reveal underlying neural representations.

• Map and uncover complex kinematic features in neuroscience research.

• Produce consistent latent spaces across various data types and experiments.

Cebra is a valuable tool for neuroscientists who wish to analyze and decode behavioural and neural data, allowing them to better understand the underlying neural representations involved in adaptive behaviours.