This is a spike neural network model of the V1, layer 5 includes Cortico-Cortical (CC), Cortico-Subcortical (CS), PV and SST neurons.
Individual simulation metrics for neuron groups are computed such as firing rates, inter-spike intervals, burst detection.
Additionally, orientation and direction selectivity are aggregated for population of cells.
.
├── data # Contains hypertuning results from amplitude simulation run
├── output # Contains output results & plots from complete simulation run (.gitignored)
├── notebooks # Jupyter notebooks used for testing different scenarios
│ ├── layer5_CC_CS_connection.ipynb # Scalar plot simulation for weighted CC->CS connection probability
│ ├── layer5_SST_Soma_selectivity.ipynb # Scalar plot simulation for weighted SST->CC/CS Soma connection probability
│ ├── layer5_sandbox.ipynb # Sandbox notebook used when creating initial network topology
│ └── ...
├── layer5_CC_CS.py # Main network file that defines the topology of the simulation
├── equations.py # Equations for different neuron types
├── parameters.py # Default parameters for simulation
├── helpers.py # Helper methods for analysis
├── plotting.py # Helper methods for plotting
├── run_amplitude_hypertuing.py # Script for running multiple simulations for amplitude hypertuning for different neurons
└── run_complete_simulation.py # Complete simulation run entrypoint
Use conda env create -n network -f requirements.txt to unpack the conda environment. Or download with conda install <module> the requirements manually.
After that, activate the environment with conda activate network
- There is only one best parameter set which I aggregated in
parameters.py. If you want to try different combinations of parameters, you can change them here. - To run main simulation
python3 run_complete_simulation.py - The project includes a script for hypertuning different input amplitudes for neuron groups. Can be run by
python3 run_amplitude_hypertuning.py - All simulation outputs are persisted in the
outputfolder. I recommend deleting the content between successive runs to avoid confusion. - Parallelization of simulations is missing. This would be a much needed improvement for the future.
- For more details and documentation about the project scope, the report can be consulted.