This is a repository to run the Fast Graph Learning for Smooth and Sparse Spectral Representation (FGL-3SR) algorithm. FGL-3SR has a significantly reduced computational complexity due to a well-chosen relaxation compared to state-of-the-art algorithms.

A demo on jupyter-notebook is available on this repo. in order to try FGL-3SR. (results obtained with cvxopt 1.1.9, numpy 1.18.5, networkx 2.3).

Learning Laplacian matrix from graph signals with sparse spectral representation. P. Humbert*, B. Le Bars*, L. Oudre, A. Kalogeratos, N. Vayatis. In the Journal of Machine Learning Research (JMLR), 22(195):1-47, 2021.

Learning laplacian matrix from bandlimited graph signals. B. Le Bars*, P. Humbert*, L. Oudre, A. Kalogeratos. In Proceedings of the International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2019.