Perceptron Ellipse

Implementation of a perceptron algorithm that trains a binary classifier to classify points inside and outside an ellipse using Python + NumPy, Numba, Matplotlib, and SciPy.

Check out the following Medium stories to know more about this project:

• Perceptron. Part 1: The Story Retold;
• Perceptron. Part 2: The Story Untold.

Installation

You need Python 3.12 and pipenv. Then, create a virtual environment and install the needed dependencies:

pipenv shell
pipenv install

Alternatively, you can use venv or any other virtual environment tool and install the following packages:

• numpy 2.0
• numba 0.60
• matplotlib 3.9
• scipy 1.14 (required for Numba to solve systems of linear equations)

Usage

Run the script __main__.py:

python .

You should see a GUI that allows you to add your own points and train the classifier. Controls:

• Left-click to add a point inside the ellipse;
• Right-click to add a point outside the ellipse;
• Enter the number of training iterations and click "Train" to train the classifier with a custom number of iterations.

Also, Matplotlib provides a toolbar with position and zoom controls.