Statistical learning approaches applied to the calculation of scaling factors for radioactive waste characterization

Biagio Zaffora (1) , Jean-Pierre Chevalier (2) , Francesco La Torre (1) , Catherine Luccioni (2) , Matteo Magistris (1) , Gilbert Saporta (3)

1 CERN [Genève]
2 CEDRIC - Centre d'études et de recherche en informatique et communications
3 CEDRIC - MSDMA - CEDRIC. Méthodes statistiques de data-mining et apprentissage

Radiological characterization is needed to dispose of the radioactive waste produced in high energy particle accelerators. We applied statistical learning methods to predict the activity of Difficult-to-Measure radionuclides-which are low-energy X , α-and β-emitters-, to establish criteria for sorting radioactive waste and to quantify prediction errors. Introduction DTM Difficult-to-Measure nuclides cannot be easily quantified by non-destructive assay means. Their activity a DTM is often correlated to the concentration of γ-emitters.

Domains: Physics [physics]; Nuclear Experiment [nucl-ex]; Statistics [stat]

Biagio Zaffora, Jean-Pierre Chevalier, Francesco La Torre, Catherine Luccioni, Matteo Magistris, et al.. Statistical learning approaches applied to the calculation of scaling factors for radioactive waste characterization. 14th Congress of the International Radiation Protection Association, May 2016, Cape Town, South Africa. ⟨10.13140/RG.2.1.3956.8240⟩. ⟨hal-02507380⟩