Easily set up a multi-layer neural network (MLP) to train on your excel datasets.
Have an easy to use application to train the MNIST dataset of hand-written digits for my pre-scientific work about neural network training.
- Train a neural network
- Test a neural network and export the output as an excel file
- Save and load (or reset) a neural networks weights and biases to be available for later training or testing
- Display the Mean Squared Error (MSE) of the neural network while in the training phase
- Select epochs and the learning rate for the neural network to gain more control on what are the ideal training settings
Requires: Python 3.7 or above
pip install --no-cache-dir -r requirements.txt
Run Source code: python main.py
Convert to executable (.exe):
pyinstaller -w --hidden-import tkinter --exclude-module PySide2 --exclude-module PyQt5 --add-data "FeedforwardNeuralNetwork.py;." main.py
- Run "main/main.exe"
The neural network is trained with the sigmoid activation function and uses the mean squared error to determine its performance on the dataset.
Project Start: 20.09.2020 (DD.MM.YYYY)
Project End: 22.09.2020 (DD.MM.YYYY)
Time spent: 06:30 (HH:MM)