Deep Learning Explorations

This project uses manim to visualize the flow of transformations of a set of 2D coordinate points through a simple neural network classifier which classifies points inside and outside of a circle. The goal is to visualize and gain an intuition for which higher-dimensional transformations a network applies to the input data in order to make the two classes linearly separable.

Inspiration for this project comes from the following sources:

1. Shape of Knowledge

2. Topology of Deep Neural Networks

3. NNs, Manifolds, Topology

Installation and Running

1. poetry install
2. poetry shell
3. python3 -m deep_learning_explorations
4. Visualizations of input and output data will be saved in the media folder.