Double Descent and Overparameterization in Particle Physics Data
Matthias Vigl, Lukas Heinrich
9/1/2025
hep-excs.LGhep-ph
No confident code implementation found for this paper.
Abstract
Recently, the benefit of heavily overparameterized models has been observed in machine learning tasks: models with enough capacity to easily cross the \emph{interpolation threshold} improve in generalization error compared to the classical bias-variance tradeoff regime. We demonstrate this behavior for the first time in particle physics data and explore when and where `double descent' appears and under which circumstances overparameterization results in a performance gain.
