Cebra

CEBRA is a machine-learning library for compressing high-dimensional time-series data—like neural recordings and behavioral videos—into low-dimensional, interpretable latent spaces. Designed for neuroscientists, computational biologists, and ML researchers, it supports both hypothesis-driven supervised and discovery-driven self-supervised modes. Key features include hybrid embedding modes, consistency metrics for comparing latent spaces, k-nearest neighbor decoding, and support for calcium imaging, electrophysiology, and behavioral data across species. The algorithm has been validated for reconstructing viewed videos from visual cortex, decoding trajectories from sensorimotor cortices, and mapping position during navigation. CEBRA integrates with Python data analysis pipelines via a scikit-learn-style API, provides built-in plotting with matplotlib and plotly, and includes GPU-accelerated training with PyTorch. Its latest extension (AISTATS 2025) adds time-series attribution maps with regularized contrastive learning. Compared to generic embedding methods like UMAP or t-SNE, CEBRA explicitly uses behavioral or auxiliary variables as labels, producing consistent, interpretable embeddings optimized for neural decoding and hypothesis testing.