Abstract
In the field of γ energy spectrum analysis, it takes a long time to obtain the measured energy spectrum with reliable characteristics, and it will inevitably bring harm to the health of the experimenters in this process. Therefore, the measured energy spectrum used for nuclide identification has a serious problem of insufficient sample size. Based on the idea of TrAdaBoost algorithm, the small amount of energy spectrum data measured by laboratory environment is used as the target sample set, and the large amount of energy spectrum data generated by Geant4 simulation is used as the auxiliary sample set. Samples useful for identification in the auxiliary sample set were transferred to the target sample set, so as to improve the nuclide identification accuracy of the target sample set. Compared with the Non-Transfer Learning method, the accuracy of identifying the target sample set nuclide type is improved by 0.87%.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have