Konstantin will work on the master’s thesis
"Investigating entropy relations in the learning process of neural networks“
When training a neural network on a given data set, some points in the data are more relevant for the final model than others. We could use this fact to select data according to their relevance to train on less points and more efficiently. My task is to figure out how to identify more important points and how to use this in data selection models. This breaks down to analyze the training process of a neural network by means of entropy and other interpretable properties.