Hyperspectral unmixing for Raman spectroscopy via physics-constrained autoencoders

Significance Hyperspectral unmixing methods are essential to exploit the capabilities of Raman spectroscopy for nondestructive, unbiased chemical characterization in a wide array of domains, from biology, chemistry, and materials to engineering and environmental science. Here, we take advantage of recent advances in machine learning and introduce a framework for Raman unmixing based on autoencoder neural networks. We demonstrate that such methods offer more versatile, robust, and data-driven Raman unmixing with improved performance compared to conventional methods in complex samples.