Shapes are not enough: CONSERVAttack and its use for finding vulnerabilities and uncertainties in machine learning applications
Authors:
- Philip BechtlePBPhilip Bechtle
- Lucie FlekLFLucie Flek
- Philipp Alexander JungPJPhilipp Alexander Jung
- Akbar KarimiAKAkbar Karimi
- Timo SaalaTSTimo Saala
- Alexander SchmidtASAlexander Schmidt
- Matthias SchottMSMatthias Schott
- Philipp SoldinPSPhilipp Soldin
- Christopher WiebuschCWChristopher Wiebusch
- Ulrich WillemsenUWUlrich Willemsen
Open Access
- 16 October 2026
- journal article
- Published by Springer Nature in The European Physical Journal C
- Vol. 86 (10) , 1172
Abstract
In High Energy Physics, as in many other fields of science, the application of machine learning techniques has been crucial in advancing our understanding of fundamental phenomena. Increasingly, deep learning models are applied to analyze both simulated and experimental data. In most experiments, a rigorous regime of testing for physically motivated systematic uncertainties is in place. The numerical evaluation of these tests for differences between the data on the one side and simulations on the other side quantifies the effect of potential sources of mismodelling on the machine learning output. In addition, thorough comparisons of marginal distributions and (linear) feature correlations between data and simulation in “control regions” are applied. However, the guidance by physical motivation, and the need to constrain comparisons to specific regions, does not guarantee that all possible sources of deviations have been accounted for. We therefore propose a new adversarial attack – the CONSERVAttack – designed to exploit the remaining space of hypothetical deviations between simulation and data after the above mentioned tests. The resulting adversarial perturbations are consistent within the uncertainty bounds — evading standard validation checks — while successfully fooling the underlying model. We further propose strategies to mitigate such vulnerabilities and argue that robustness to adversarial effects must be considered when interpreting results from deep learning in particle physics.Keywords
Funding Information
- Bundesministerium für Forschung, Technologie und Raumfahrt (05D23UM1)
- Deutsche Forschungsgemeinschaft (3037/533607693)
This publication has 15 references indexed in Scilit:
- A Survey of Large Language ModelsFrontiers of Computer Science, 2026
- Enforcing Fundamental Relations via Adversarial Attacks on Input Parameter CorrelationsComputing and Software for Big Science, 2025
- How Deep Learning Sees the World: A Survey on Adversarial Attacks & DefensesIEEE Access, 2024
- Adversarial attacks and defenses in explainable artificial intelligence: A surveyInformation Fusion, 2024
- Recent Advancements in End-to-End Autonomous Driving Using Deep Learning: A SurveyIEEE Transactions on Intelligent Vehicles, 2023
- Advances in Adversarial Attacks and Defenses in Computer Vision: A SurveyIEEE Access, 2021
- Deep learning for healthcare: review, opportunities and challengesBriefings in Bioinformatics, 2017
- Observation of a new boson at a mass of 125 GeV with the CMS experiment at the LHCPhysics Letters B, 2012
- Observation of a new particle in the search for the Standard Model Higgs boson with the ATLAS detector at the LHCPhysics Letters B, 2012
- Asymptotic formulae for likelihood-based tests of new physicsThe European Physical Journal C, 2011