Machine learning study on nuclear $\alpha$ decays

by 민수 권 (MS Candidate)

Friday May 14, 2021, 3:00 PM → 3:30 PM Asia/Seoul

206호 (제1과학관)

The regression process of machine learning is applied to predict the half-lives of nuclear alpha decays. The artificial neural networks are used as the models of this thesis. The predictive power of the artificial neural networks using 164 experimental data is investigated in this thesis with two approaches. In ModelI, we trained the artificial neural networks to experimental data of nuclear alpha decays half-lives directly. Proton numbers, neutron numbers, and Qα values are also included in this training. In ModelII, we trained the artificial neural networks to the difference be- tween the experimental data and the theoretical model, Viola-Seaborg formula. The purpose of building ModelI is to verify the applicability of machine learning to nu- clear alpha decays and the purpose of building ModelII is to apply the technique to estimate uncertainties of theoretical models. The results of this thesis show that there are possibilities of improving the predictive power of empirical models by using machine learning techniques.

Thesis Advisor: Prof. Yongseok Oh

• URI http://www.dcollection.net/handler/knu/000000100381
• UCI I804:22001-000000100381