[3I09-13] Theoretical Analysis and Surrogate Reaction

[3I10] Estimation of nuclide production cross sections through transfer learning

*Hiroki Iwamoto¹, Shin-ichiro Meigo¹, Kenta Sugihara² (1. JAEA, 2. KEK)

Keywords:nuclear data, nuclide production cross section, machine learning, transfer learning, Gaussian process

A model has been developed to estimate nuclide production cross sections by transfer learning using Gaussian processes. The learning model was found to reproduce experimental data for a wide range of incident energies and target elements, and to produce estimation results similar to the physics model even in regions where experimental data are lacking.