Iterative reconstruction of signals on graph

TL;DR

This paper introduces an iterative algorithm for reconstructing signals on graphs from limited samples, improving convergence and efficiency over existing methods.

Contribution

It adapts the Papoulis-Gerchberg algorithm for graph signals, optimizing a key constant to enhance performance.

Findings

Achieves comparable or better reconstruction accuracy

Converges faster than existing algorithms

More computationally efficient

Abstract

We propose an iterative algorithm to interpolate graph signals from only a partial set of samples. Our method is derived from the well known Papoulis-Gerchberg algorithm by considering the optimal value of a constant involved in the iteration step. Compared with existing graph signal reconstruction algorithms, the proposed method achieves similar or better performance both in terms of convergence rate and computational efficiency.