Note
The documentation provided here is currently under construction. Please also visit the slack channel #event-generator for further questions.
About the Project
The Event-Generator project is a software framework that enables the reconstruction of arbitrary IceCube events via a hybrid method composed of Maximum Likelihood Estimation and Deep Learning.
Neural networks (NNs) are used to model the high-dimensional and complex relation between an
event hypothesis and expected light yield in the detector. Once these NNs are trained and
exported, they can be used in a typical maximum likelihood setting via the provided
I3TraySegments (Apply Exported Models).
Further Material & Publications