Haverson, K. C. Z. and Smith, R. and Shenfield, A. and Fortulan, R. (2025) A ResNeXt50-based convolution neural network for nuclear reaction classification in an active target TPC detector. Il nuovo cimento C, 48 (1). pp. 1-8. ISSN 1826-9885
|
Text
ncc13189.pdf - Published Version Download (283kB) | Preview |
Abstract
The Convolutional Neural Network (CNN), ResNeXt50, was used to classify nuclear reactions measured in a gas-filled charged particle detector. This active-target time-projection chamber, built by the University of Warsaw, is optimised for studying photo-dissociation reactions using intense γ-beams. In particular, the 16O(γ,α)12Cand12C(γ,3α) reactions were measured using γ-beams at the HIγS facility at Duke University. Reactions of interest were distinguished from other background channels using the CNN. For this preliminary study, a small sample of two hundred hand-classified events per category were used to train the model. This was then used to classify an unseen data set comprising one hundred events in each category. This method was extremely effective at removing background events and could differentiate the two main reaction channels with 96% accuracy.
| Item Type: | Article |
|---|---|
| Subjects: | 500 Scienze naturali e Matematica > 530 Fisica |
| Depositing User: | Marina Spanti |
| Date Deposited: | 24 Mar 2025 15:20 |
| Last Modified: | 24 Mar 2025 15:20 |
| URI: | http://eprints.bice.rm.cnr.it/id/eprint/23500 |
Actions (login required)
 |
View Item |
