Designing Observables for Measurements with Deep Learning
arXiv:2310.08717 · doi:10.1140/epjc/s10052-024-13135-4
Abstract
Many analyses in particle and nuclear physics use simulations to infer fundamental, effective, or phenomenological parameters of the underlying physics models. When the inference is performed with unfolded cross sections, the observables are designed using physics intuition and heuristics. We propose to design targeted observables with machine learning. Unfolded, differential cross sections in a neural network output contain the most information about parameters of interest and can be well-measured by construction. The networks are trained using a custom loss function that rewards outputs that are sensitive to the parameter(s) of interest while simultaneously penalizing outputs that are different between particle-level and detector-level (to minimize detector distortions). We demonstrate this idea in simulation using two physics models for inclusive measurements in deep inelastic scattering. We find that the new approach is more sensitive than classical observables at distinguishing the two models and also has a reduced unfolding uncertainty due to the reduced detector distortions.
This is the version published in EPJC
References in corpus (35)
- New CTEQ global analysis of quantum chromodynamics with high-precision data from the LHC
- The Path to Proton Structure at One-Percent Accuracy
- OmniFold: A Method to Simultaneously Unfold All Observables
- A generic anti-QCD jet tagger
- Mining gold from implicit models to improve likelihood-free inference
- Constraining Effective Field Theories with Machine Learning
- Adversarially-trained autoencoders for robust unsupervised new physics searches
- TUnfold: an algorithm for correcting migration effects in high energy physics
- A Guide to Constraining Effective Field Theories with Machine Learning
- Decorrelated Jet Substructure Tagging using Adversarial Neural Networks
- Invertible Networks or Partons to Detector and Back Again
- Neural Networks for Full Phase-space Reweighting and Parameter Tuning
- Thinking outside the ROCs: Designing Decorrelated Taggers (DDT) for jet substructure
- How to GAN away Detector Effects
- DisCo Fever: Robust Networks Through Distance Correlation
- Machine Learning Uncertainties with Adversarial Neural Networks
- ABCDisCo: Automating the ABCD Method with Machine Learning
- A deep neural network to search for new long-lived particles decaying to jets
- Uncertainty Aware Learning for High Energy Physics
- INFERNO: Inference-Aware Neural Optimisation
- uBoost: A boosting method for producing uniform selection efficiencies from multivariate classifiers
- Presenting Unbinned Differential Cross Section Results
- Mass Unspecific Supervised Tagging (MUST) for boosted jets
- Convolved Substructure: Analytically Decorrelating Jet Substructure Observables
- Mass Agnostic Jet Taggers
- Reconstructing the Kinematics of Deep Inelastic Scattering with Deep Learning
- Disentangling Quarks and Gluons with CMS Open Data
- Reducing the dependence of the neural network function to systematic uncertainties in the input space
- Measurement of the differential production cross section as a function of the jet mass and extraction of the top quark mass in hadronic decays of boosted top quarks
- Unbinned multivariate observables for global SMEFT analyses from machine learning
- Optimal statistical inference in the presence of systematic uncertainties using neural network optimization based on binned Poisson likelihoods with nuisance parameters
- Unbinned Profiled Unfolding
- neos: End-to-End-Optimised Summary Statistics for High Energy Physics
- Enhancing searches for resonances with machine learning and moment decomposition
- Preservation through modernisation: The software of the H1 experiment at HERA