Forward Models in the Cerebellum using Reservoirs and Perturbation Learning

Code for training a forward model as described in Schmid, K., Vitay, J. and Hamker, F. (2019). "Forward Models in the Cerebellum using Reservoirs and Perturbation Learning". CCN 2019.

The reservoir is based on the model in Roessert, C., Dean P. and Porrill, J. (2015). “At the edge of chaos: how cerebellar granular layer network dynamics can provide the basis for temporal filters.” PLoS computational biology, 11(10). doi.org/10.1371/journal.pcbi.1004515.

Perturbation learning is based on the algorithm described in Bouvier, G. et al. (2018). “Cerebellar learning using perturbations.” eLife 7. doi.org/10.7554/eLife.31599.

Dependencies

• ANNarchy - a neural simulator
• numpy - package for scientific computing with Python
• matplotlib - 2D plotting library

Run Simulations

To train and test the network on the provided data, execute the following command:

python start.py training_and_test_sets.npz <output directory>

Results will be saved in npz-format to <output directory>.

Plot Results

To plot the results, run:

python plot_mean_error.py <results.npz>
python plot_responses.py <results.npz>
python plot_trajectory.py <results.npz>
python plot_pn_activity.py <results.npz>