Neural Networks in Numpy

The following repository contains code for training neural networks only in numpy. It is a attempt to encourage an encourage a clear understanding of how neural networks are trained : Backprop and Gradient Descent

FC Network - Regression

---

Function_Regression.ipynb The notebook contains code to train a neural network to overfit any function in a given range

FC Network - Classification

---

Classification_MNIST.ipynb The notebook aims at training a neural network to classify the digits in the MNIST dataset

TODO

---

• Adam Optimizer
• CNN in numpy
• RNN in numpy