graph uncertainty principle calculation in multi-dimensional feature case
cjfcsjt opened this issue · comments
Hi! Thanks for your great work! I have a question about the calculation of the graph uncertainty princple. If each node in a graph corresponds to a mutli-dimensional feature vector, how to measure the concentration in vertex and frequency domain? Should I change the vector p-norm || f ||_p * || f_hat ||_p to the matrix p_norm?
Hi, I guess you are right. If I understand you correctly you want to extend measurements of space-frequency concentrations to multivariate signals on graphs. There is certainly more than one way to achieve this. If you are measuring the space-frequency concentrations separately for each component of your multi-dimensional vector and then you are combining them in form of a mean value you should end up with a particular kind of matrix norm for your multivariate feature vector. So, I'm pretty sure your suggestion is a valuable approach.