simonmichau / dynamic-memory-traces-for-sequence-learning-in-spiking-network

Repo of the bachelor thesis 'Dynamic memory traces for sequence learning in spiking networks'

Info: This is the development version of the project. For a cleaned up and more usable version check out the sequence-learning repository.

Dynamic memory traces for sequence learning in spiking networks

Repo of the bachelor thesis 'Dynamic memory traces for sequence learning in spiking networks'

Setup

To run the code from the nest/ subdirectory NEST and NESTML (with gh pr checkout 805 for issue #805) are required.

To run the code in the legacy/ subdirectory Python 2.7 is required.

Changing the NESTML

Updating the custom NESTML neuron and synapse models can be a bit tedious. The easiest way to do so is laid out here:

1. Generate a new target by setting the global variables NEURON_MODEL and SYNAPSE_MODEL to the names of your custom models in the nest/nestml_models subdirectory and calling python nest_network.py regen_models. This will create and install a new target in the nest/nestml_targets subdirectory.
2. Reimport all the custom changes made to support normalization_sum and InstantaneousRateConnectionEvent in the corresponding .cpp and .h files.
3. Run make -j 4 install to recompile the model with the custom changes.