Research on the topic of uncertainty prediction, using bayesian approaches, calibration and popular methods like MC Dropout or Natural Gradient
This github repo contains the various code files and results in the form of plots that I have used for my research into calibration and uncertainty prediction. The code is (mostly) uncommented as it is for personal use. The plots are saved in each of the associated folders. For an exhaustive presentation of the research and a better understanding of the codes and plots, please refer to the pdf report, which covers all the research carried out during the course and explains it in detail (please note that it is written in French).