Krzysztof Graczyk

Bayesian framework, neural networks and nucleon form factors

The review of the Bayesian approach for feed-forward neural networks will be given. The aim is to introduce the basis of the Bayesian statistics and then to show how to adopt this framework for feed-forward neural networks. As an application I will show the extraction of the electromagnetic proton form factors and two photon exchange correction from the elastic ep scattering data. The seminar will have rather informal character.