Neural networks from scratch

Project for the Machine Learning course @ University of Pisa

Overview

1. Short description
2. Directory structure
3. Quick start

---

Description

This project contains the implementation from scratch of neural networks for classification and regression trained with Stochastic Gradient Descent with back-propagation.
For more detailed information check the report

Directory structure (only main elements)

ML-project
  │── src
  │    │── network.py
  │    │── layer.py
  │    │── functions.py
  │    │── optimizers.py
  │    │── model_selection.py
  │    │── weights_initializations.py
  │    └── demo.py
  └── datasets
  │    └── cup
  │    │   │── ML-CUP20-TR.csv
  │    │   │── ML-CUP20_TS.csv
  │    │   │── CUP-DEV-SET.csv
  │    │   └── CUP-INTERNAL-TEST.csv
  │    └── monks
  │        │── monks.names       # description file
  │        │── monks-x.train     # the 'x' is the number of the dataset (1, 2, 3)
  │        └── monks-x.test
  └── plots
  │   │── ensemble      # where the plots of the constituent models go
  │   └── monks         # where the plots of the monks are
  └── results           # json files with the results of grid searches
  └── ensemble_models   # json files with the constituent models of the ensemble

Quick start

Install Python:
sudo apt install python3

Install pip:
sudo apt install --upgrade python3-pip

Install requirements:
python -m pip install --requirement requirements.txt

MONKS demo: open the script to find information and instructions. Then execute it with

cd src/
python demo.py