Elvin Isufi

Papers

Total Citations

H-Index

About

Elvin Isufi is a leading researcher in the field of graph signal processing and geometric deep learning, with a primary focus on the stability and robustness of graph neural networks (GNNs). His work bridges the gap between classical signal processing on graphs and modern deep learning architectures, particularly by analyzing how graph convolutional neural networks (GCNNs) behave under topological changes. Isufi’s major contributions include a rigorous theoretical framework for understanding the stability of GCNNs to stochastic perturbations in graph structure—a critical issue for deploying these models in real-world, dynamic networks where the underlying topology may be noisy or time-varying. His most-cited paper, “Stability of graph convolutional neural networks to stochastic perturbations” (2021), has garnered 23 citations and provides foundational insights into how random edge failures or additions affect learned representations. Isufi has also authored influential works such as “Graphs, Convolutions, and Neural Networks” and “From Graph Filters to Graph Neural Networks” (2020), each with 12–15 citations, which systematically connect classical graph filter theory to modern GNN design. His research is particularly impactful for applications in sensor networks, social network analysis, and infrastructure monitoring, where robustness to structural uncertainty is paramount.