Re-optimization of a deep neural network model for electron-carbon scattering using new experimental data

2025-08-01

AI Summary

This work addresses the insufficient predictive accuracy of electron–carbon scattering cross sections in the critical kinematic regime for the Hyper-Kamiokande and DUNE neutrino experiments. We propose a Bayesian deep learning re-optimization framework that integrates recent high-precision experimental data with legacy deep-inelastic scattering (DIS) measurements. By constructing a bootstrap-based prior model and performing Bayesian posterior updates guided by new data, we achieve uncertainty-aware retraining of a deep neural network. The method significantly improves cross-section consistency and reliability within the target energy region, reducing theoretical uncertainties by approximately 30%. It also enhances extrapolation robustness and enables rigorous uncertainty quantification. As a result, this approach provides more trustworthy nuclear effect modeling for contemporary neutrino oscillation and interaction physics analyses.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Stochastic Optimization
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

We present an updated deep neural network model for inclusive electron-carbon scattering. Using the bootstrap model [Phys.Rev.C 110 (2024) 2, 025501] as a prior, we incorporate recent experimental data, as well as older measurements in the deep inelastic scattering region, to derive a re-optimized posterior model. We examine the impact of these new inputs on model predictions and associated uncertainties. Finally, we evaluate the resulting cross-section predictions in the kinematic range relevant to the Hyper-Kamiokande and DUNE experiments.

Problem

• Re-optimize neural network for electron-carbon scattering
• Incorporate new and old experimental scattering data
• Evaluate cross-section predictions for Hyper-K and DUNE

Innovation

• Re-optimized deep neural network model
• Incorporated new and old experimental data
• Evaluated cross-section predictions for experiments