Deep calibration transfer: Transferring deep learning models between infrared spectroscopy instruments

• A tranfer learning based approach to transfer deep calibrations. The apporach was demonstrated on two cases: tablets and olives. Data augmention used develop primary models. successfully transferred models between instrument. Calibration (CT) is required when a model developed one instrument needs be new Several methods are available in the chemometrics domain multivariate calibrations using modelling techniques such as partial least-square regression. However, recently (DL) gaining popularity spectral data. traditional CT not suitable which neural networks architectures. Hence, this study presents concept of calibration for transferring DL made onto from domain. To show it, different cases presented. first case benchtop FT-NIR (Fourier Transform Near Infrared) instruments, second handheld NIR (Near instruments. In both cases, performed standard-free i.e., no common standard samples were estimate any function. results showed that with CT, can easily adapted main benefit it free does require sample measurements. Such instruments support widespread sharing chemometric scientific practitioners.