In this project we do linear regression on the Franke function and terrain data. Run code using "python main.py", configure parameter file "paramfile.py" for each run. Several example parameter settings can be found in the file for reproduction of article data. Also included are parameter-settings for testing the predictions agianst SKLEARN, use Test=True.
In this project we attempt to predict the probability of default from credit card data. We apply both logistic regression and neural networks for this. Run code using "python part#.py".
Part1.py - Running the logistic regression method and comparing to sklearn.
Part2.py - Running the feed-forward neural network on credit card data.
Part3.py - Running the feed-forward neural network on the franke function data.
In this project we implement a convolutional neural network manually without the help of keras, pytorch or tensorflow etc. The aim of the project is to analyse how the addition of a convolutional layer can help classification compared to a pure dense neural network.
The network is applied to MNIST, CIFAR and SVHN.