Team Members: Sophanna Ek, Melissa Hazlewood, Rowan Herbert, Dennis La
ml_perceptron/
src/
data.py
- Contains methods for retrieving the vectors needed for the training and validation of the program.
main.py
- The entry point to the program.
mlp.py
- Contains the multi layer perceptron
mlp
class and handles its instantiation as well as forward and backward propogation of training vectors and error respectively.
- Contains the multi layer perceptron
res/
data.txt
- Contains the raw data vectors.
- A basic understanding of programming and terminal emulator know how.
- An installation of
Python3
(orPython
) which can be found here. - A Python package manager such as pip. It can be installed by following the instructions here.
- The NumPy package for Python. It can be installed with
pip
by running the terminal commandpip install numpy
.
Cloning from GitHub:
- You can
Git
clone the GitHub repo withgit clone https://github.com/RLHerbert/ml_perceptron.git
.
Follow these steps to run the program:
- Navigate to the
ml_perceptron
folder in your terminal emulator of choice. - Run Python3 by typing and entering
Python3 src/main.py
in your terminal.- Alternatively, if you wish to save the resulting output you may enter
Python3 src/main.py > [output file]
on Linux/MacOS terminals and Windows PowerShell.
- Alternatively, if you wish to save the resulting output you may enter
This project features a multilayer perceptron classifier which utilizes forward propogation training and backward propogation error correction to classify the provided data set. The program shows the initial and final hidden and output weights. It also outputs the precision, recall, sensitivity and specificity of a given class and the accuracy and error rates of the MLP in general.