Skip to content

Latest commit

 

History

History
26 lines (16 loc) · 1.52 KB

README.md

File metadata and controls

26 lines (16 loc) · 1.52 KB
Raw

Firing Rate Model of V1, Layer 5

This model of the V1, layer 5 includes Cortico-Cortical (CC), Cortico-Subcortical (CS), PV and SST neurons. Plasticity is not implemented yet. In this model, we observe orientation and direction selectivity.

Requirements

Use conda env create -f network.yml to unpack the conda environment. Or download with conda install <module> the requirements manually. After that, activate the environment with conda activate network

Run simulation

Use python3 ds_os_model.py <put config here> where the second argument is the config you want to use. For instance use configs.test_config

Some hints

  • the file ds_os_model is used to find suitable hyperparameters. If you want to just run one simulation delete the last parallelization part
  • the model was implemented to measure direction and orientation selectivity. You might want to change the input (in Implementation/helper.py) and evaluation metrics according to your project
  • it's advised to plot the activity over time to check the simulation

Output

  • The output both as figures and csv file. Figures are listed as learning_rule-simulation_number_act/weight.png
  • The csv file concludes the spatial variance, time variance and the mean final expression value of the system. All column are post-pre synpatic neurons.

Parameter tuning

  • Current method: Bayesian parameter inference. (Potentially varational bayesian inference in the future)
  • Distribution input and distribution output surrounding the real value.