This repository is for a project that involves the construction of a basic neural network (MLP), and the performance comparison against the commonly used libraries (Py Torch and Tensorflow Keras).
If you aren't familiar with neural networks, go to the 'Artificial_Neural_Networks' file, where I explain the basics concepts with a MLP. If you're not familiar with Machine Learning either, you should learn it first in order to understand how a Deep learning model works.
In order to run all the codes succesfully, you'll need to create a python virtual environment and install all the necessary requirements.
- Create the virtual environment (there is no mandatory version, but
python 3.9
is recommended)python -m venv <venv_name>
- Activate the environment
Whenever you want to get out from the environment, just run the next command:
source <venv_name>/bin/activate
deactivate
- Install the necessary libraries through the 'Makefile'
pip install -r requirements.txt
Once you are all set, feel free to browse through the project.
- Go to 'src' to see the MLP class and its functions
- Go to 'test' to see the functions used for the unit tests, and try to test the code yourself
pytest tests/test_neural_network.py
- Go to the Jupyter Notebooks which contain the examples, and run them yourself
- Keep in mind that this project is very basic because the purpose is purely educational, the ANN created won't work properly in a project with production standards
- You can try with your own data sets (just be careful to do the right preprocessing), and with other configurations and topologies