Score: 0

Double Descent and Overparameterization in Particle Physics Data

Published: September 1, 2025 | arXiv ID: 2509.01397v1

By: Matthias Vigl, Lukas Heinrich

Potential Business Impact:

Makes computer models better at guessing physics results.

Business Areas:
A/B Testing Data and Analytics

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.

Country of Origin
🇩🇪 Germany

Page Count
6 pages

Category
Physics:
High Energy Physics - Experiment