Towards A Theory of Duality for Graph Signal Processing
share
Motivated by duality in traditional signal processing, we investigate the concept of duality for graph Fourier transforms. Given two graphs, we define their dualness, a measure of how close the graphs are to being (signal processing) duals of each other. We show that this definition satisfies some desirable properties, and develop an algorithm based on co-ordinate descent and perfect matching to compute the dualness when the graphs have distinct eigen values.
READ FULL TEXT