This joint work with Prof. Tirza Routtenberg will be presented at ICASSP 2025, the leading conference in signal processing.
? This work is also published as a journal paper in the IEEE Transactions on Signal Processing. https://ieeexplore.ieee.org/document/10750048
? In this work, we explored sparse recovery when the dictionary matrix is associated with an underlying graph. Our key contribution is establishing a link between the orthogonality of dictionary atoms—each corresponding to a graph node—and the spatial arrangement of nodes on the graph.
Building on this, we developed three methods offering different precision-complexity trade-offs.