Demo Notebooks#

We provide a set of demo notebooks to get started with using CEBRA. To run the notebooks, you need a working Jupyter notebook server, a CEBRA installation, and the datasets required to run the notebooks, available on FigShare.

Getting Started with CEBRA
Encoding of space, hippocampus (CA1)
Decoding movie features from (V1) visual cortex
Forelimb dynamics, somatosensory (S1)
Synthetic neural benchmarking
Hypothesis-driven analysis
Consistency
Decoding
Topological data analysis
Unified encoders via behavioral alignment
Technical: Training models across animals
Technical: conv-piVAE
Technical: S1 training with MSE loss
Technical: Learning the temperature parameter
Demo: Using OpenScope Data
Demo: Using Dandi Data
Explainability: xCEBRA on RatInABox dataset

The demo notebooks can also be found on GitHub.

Installation#

Before you can run these notebooks, you must have a working installation of CEBRA. Please see the dedicated Installation Guide for information on installation options using conda, pip and docker.

Synthetic Experiment Demo (CEBRA, piVAE, tSNE, UMAP): This demo requires several additional packages that have differing requirements to CEBRA. Therefore, we recommend using the supplied docker container or conda cebra-full env.

Demo Data#

We host prepackaged data on figshare. And several of the demo notebooks have an automatic data download function.

If you don’t see the auto-download, and you use Google Colaboratory, you can easily add the following code into an early cell in the notebook to directly download and use:

#for google colab only, run this cell to download and extract data:
!wget --content-disposition https://figshare.com/ndownloader/files/36869049?private_link=60adb075234c2cc51fa3
!mkdir data
!tar -xvf "/content/data.tgz" -C "/content/data"

For different paths, you can specify the CEBRA_DATADIR=... environment variable. You can do this by placing import os; os.environ['CEBRA_DATADIR'] = "path/to/your/data" at the top of your notebook.

Contributing#

We welcome Demo notebooks from others! Please fork the repo, add your notebook, check that it works on Google Colaboratory (remove the launch button within your PR), and then open a PR! Please also edit the “gallery” list (see the soure code for this page), and finally, add an icon here, then a path to the icon here.

For reference, the original open-source data we used in Schneider, Lee, Mathis 2023 is available at: