/ANNFromScratch

Building a Neural Network from scratch using Tensorflow

Primary LanguageJupyter NotebookMIT LicenseMIT

Artificial Neural Network from Scratch

In this interactive Notebook we are going to:

  • Build the architecture of a simple 2-layers ANN
  • Generate data
  • Train the network
  • Test the network and evaluate metrics.

All the tasks are done using low-level Tensorflow API, to highlight the math beyond ANNs.

The notebook provides explanations and schemes to better understand all steps.
Just make sure you have Python 3.6 or higher, Jupyter and all the required dependencies
listed in dependencies.txt. The first paragraph will tell you if the needed libraries are not yet installed.

Play with controls and assign new values to Hyperparameters to see what happens to the Network Performace.