We investigate the importance of small singular values and present a theoretical model that explains their relevance.

Learning spectral representation from diverse hyperspectral data to improve plant trait prediction (multi-regression)