Simulation-Prior Independent Neural Unfolding Procedure
This addresses the challenge of unbiased data unfolding in high-energy physics, offering a domain-specific improvement over traditional methods.
The paper tackles the problem of unfolding high-dimensional data without binning at the LHC by introducing the SPINUP method, which extracts unfolded distributions independently of simulation priors using neural networks and neural importance sampling, and demonstrates it on jet substructure observables and Higgs and single-top production with parton-level unfolding.
Machine learning allows unfolding high-dimensional spaces without binning at the LHC. The new SPINUP method extracts the unfolded distribution based on a neural network encoding the forward mapping, making it independent of the prior from the simulated training data. It is made efficient through neural importance sampling, and ensembling can be used to estimate the effect of information loss in the forward process. We showcase SPINUP for unfolding detector effects on jet substructure observables and for unfolding to parton level of associated Higgs and single-top production.