`vonNeumann()` computes the von Neumann entropy of a kernel matrix. Entropy values close to 0 indicate that all its elements are very similar, which may result in underfitting when training a prediction model. Instead, values close to 1 indicate a high variability which may produce overfitting.
References
Belanche-Muñoz, L.A. and Wiejacha, M. (2023) Analysis of Kernel Matrices via the von Neumann Entropy and Its Relation to RVM Performances. Entropy, 25, 154. doi:10.3390/e25010154. Link