/Handmade-MLP

A multi-layer perceptron implementation only using numpy.

Primary LanguagePython

Handmade-MLP

A multi-layer perceptron implementation only using numpy.

We implement MLP, MLQP(Quadratic Perceptron) and their backward propagation algorithms based on MSE using numpy. Then we apply it to classify points on double helix.

File Structure

  • network.py: MLP and MLQP implementation
  • main.py: main-file
  • data: coordinates of points on double helix
  • res: training loss and accuracy