The article discusses the use of Raman Spectroscopy in biotechnology as a process analytical technology. It focuses on the use of Convolutional Neural Networks (CNNs) for handling non-Gaussian noise and accounting for beam misalignment, pixel malfunctions, or the presence of additional substances. The paper emphasizes the need for large amounts of training data for CNNs and discusses how data augmentation can improve the generalization of the neural network. The study demonstrates the method using synthetic spectra of Ralstonia eutropha batch cultivations.
Publication date: 1 Feb 2024
Project Page: https://arxiv.org/abs/2402.00851v1
Paper: https://arxiv.org/pdf/2402.00851
