paulbfb / neural_networks_functions Goto Github PK

This Git repository is primarily concerned with the evaluation of different loss functions in the framework of neural networks.

First, the file named “Loss Functions for Machine Learning Algorithms” gives a short theoretical introduction to the most common loss functions used in neural networks, and how these relate to the overarching framework of maximum likelihood estimation.

Second, the jupyter notebook named “loss_functions_graphical” implements several different loss functions and plots the results to visualize the differences.

This repository was created by Paul Leitner and Matthias Sammer, whereas Mr. Leitner was in charge of implementation and Mr. Sammer provided the theoretical background.