This repository folder is for Project 2 in FYS-STK4155 – Anvendt dataanalyse og maskinlæring, for project group 40. 
The group members are Yoan Tutunarov, Sebastian Koranda and Vlasislav Foss

Short description of the project
Machine learning is a subset of AI that focuses on algorithms that can analyse data and make predictions. One of the most influential 
algorithms in ML is Neural Networks. In the following study, we will implement our own Neural Network and dive into its architecture.
Thus, we will have a look at the mathematical foundations of gradient-based optimization, activation functions of the different layers, 
different loss functions and bias-variance tradeoff.
Similarly to Project 1, we will use the Runge function when developing our neural network code for regression problems. We will later 
discuss how our NN performs compared to OLS used in Project 1. 
In addition, the MNIST (digit recognition dataset) will be used for multiclass classification task. 
Through out the study, we will carry out comparisons between our NN code and similar code using Sci-kit learn, and between our calculated
derivatives and those by automatic differentiation as Autograd library. 

The required packages that have to be installed for running the codes are: 
numpy -> installation guide:  https://numpy.org/install/
skikit-learn -> installation guide:  https://scikit-learn.org/stable/install.html
matplotlib -> installation guide:  https://matplotlib.org/stable/install/index.html
autograd -> installation guide:  https://autograd.readthedocs.io/en/latest/installation.html

The notebooks with the various codes are stored in subfolders and files inside project2 folde. For example, cost functions have 
a separate file for each cost function inside the folder named "cost". The same for activation functions, data, figs (figures) and 
optimizers, The codes for methods as NN, regression, clasification and gradients are in each their own file inside the main 
folder (project2). You will also find the repport and the project description in the same folder.