Double Descent and Overparameterization in Particle Physics Data

2025/09/01 by Matthias Vigl, L. Heinrich, Vigl, Matthias +1 · 1 voice
Computer Science · Physics and Astronomy · #Data Analysis #FOS: Computer and information sciences #FOS: Physical sciences #Gaussian Processes and Bayesian Inference #High Energy Physics - Experiment (hep-ex) #High Energy Physics - Phenomenology (hep-ph) #Machine Learning (cs.LG) #Machine Learning and Data Classification #Particle physics theoretical and experimental studies #Statistics and Probability (physics.data-an)

paper · pdf · doi:10.48550/arxiv.2509.01397

openalex publication_date 2025/09/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Recently, the benefit of heavily overparameterized models has been observed in machine learning tasks: models with enough capacity to easily cross the 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.

Citations

• Transforming jet flavour tagging at ATLAS
• Towards Foundation Models for Experimental Readout Systems Combining Discrete and Continuous Data
• Double descent in quantum kernel methods
• OmniJet-α_C: Learning point cloud calorimeter simulations using generative transformers
• Is Tokenization Needed for Masked Particle Modelling?
• Finetuning Foundation Models for Joint Analysis Optimization
• Masked Particle Modeling on Sets: Towards Self-Supervised High Energy Physics Foundation Models
• When and how epochwise double descent happens
• Deep Double Descent: Where Bigger Models and More Data Hurt
• The Implicit Bias of Gradient Descent on Separable Data

Discussions

• 1/New paper led by Matthias Vigl arxiv.org/abs/2509.01397 - we wanted to see whether you can see double descent and a benefit of overparametrization in particle physics data and tasks. We see both mod [bsky, 8 points, 1 comments]

Related

• A Farewell to the Bias-Variance Tradeoff? An Overview of the Theory of Overparameterized Machine Learning
• Reconciling modern machine learning practice and the bias-variance trade-off
• A U-turn on Double Descent: Rethinking Parameter Counting in Statistical Learning
• Understanding the Double Descent Phenomenon in Deep Learning
• Understanding the double descent curve in Machine Learning