Self-supervised learning for inter-laboratory variation minimization in surface-enhanced Raman scattering spectroscopy

). The leave-one-lab-out cross-validation (LOLABO-CV) results indicate that the MVNet also minimizes the variance of completely unseen laboratory datasets while improving the reproducibility and linear fit of the regression model. The Python implementation of MVNet and the code for the analysis can be found on the GitHub page https://github.com/psychemistz/MVNet.