Electron-nucleus cross sections from transfer learning

TL;DR

This paper demonstrates that transfer learning enables a neural network trained on electron-carbon scattering data to accurately predict electron-nucleus cross sections across various nuclear targets, improving efficiency in nuclear physics modeling.

Contribution

The study introduces applying transfer learning to physics, specifically for predicting electron-nucleus cross sections, which is a novel approach in this domain.

Findings

Transfer learning improves prediction accuracy for different nuclei.

Fine-tuned neural networks match experimental cross section data.

Method reduces need for extensive new training data.

Abstract

Transfer learning (TL) allows a deep neural network (DNN) trained on one type of data to be adapted for new problems with limited information. We propose to use the TL technique in physics. The DNN learns the details of one process, and after fine-tuning, it makes predictions for related processes. We consider the DNNs, trained on inclusive electron-carbon scattering data, and show that after fine-tuning, they accurately predict cross sections for electron interactions with nuclear targets ranging from helium-3 to iron.