Neural Networks
Function approximation, MNIST classification and denoising
A course project, comprising three objectives.
In the first part, points are generated for a function f(x) and neural networks are explored as a means to approximate this function.
The second part included classifying MNIST and examining the effects of different parameters.
Finally, NNs are used as a means of de-noising samples of MNIST (to which gaussian and salt noise was added) and this method is explored thoroughly.
The repository contains code and more explanations.