Analysis of the variances of the weights in Hebbian learning with Oja's rule. For being able to use this repository please create a virtual environment with all requirements installed.
python -m venv venv
pip install -r requirements.txt
Once you have the environment you can run python oja.py
to create JSON files with the variance values for different correlations and learning rates saved.
You can turn these JSON files into a single plot saved as a PNG file using python visualize.py
.