Getting started: Overview & Installation¶

Overview¶

This is a small neural network framework that has been created from scratch. It relies mainly on NumPy for efficient operations on n-dimensional arrays. Being a team of only 4 developers, we restricted the functionality to the following neural network architectures:

• Multi-Layer Perceptrons (MLP)

• (Denoising) Autoencoders (AE)

• Variational Autoencoders (VAE)

• Generative Adversarial Networks (GAN)

Hence in the next pages of this online documentation, we will briefly recapitulate the basics of deep learning and show how the fundamental building blocks of Multi-Layer Perceptrons have been implemented in our framework. Then we will continue with a few demos of the afore-mentioned neural network architectures and give some insights in the theory behind these models.

Installation guide¶

Clone repository:

git clone https://github.com/mathmerizing/Autoencoder.git

pip install -r requirements.txt