BAYESIAN NODE CLASSIFICATION FOR NOISY GRAPHS
| Autor: | Hakim Hafidi, Philippe Ciblat, Mounir Ghogho, Ananthram Swami |
|---|---|
| Přispěvatelé: | CIBLAT, Philippe, Université Internationale de Rabat (UIR), Laboratoire Traitement et Communication de l'Information (LTCI), Institut Mines-Télécom [Paris] (IMT)-Télécom Paris, Département Communications & Electronique (COMELEC), Télécom ParisTech, Communications Numériques (COMNUM), Institut Mines-Télécom [Paris] (IMT)-Télécom Paris-Institut Mines-Télécom [Paris] (IMT)-Télécom Paris, Institut Polytechnique de Paris (IP Paris) |
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
Computer science
[INFO.INFO-TS] Computer Science [cs]/Signal and Image Processing Bayesian probability 02 engineering and technology 01 natural sciences [INFO.INFO-NI]Computer Science [cs]/Networking and Internet Architecture [cs.NI] [MATH.MATH-IT] Mathematics [math]/Information Theory [math.IT] [INFO.INFO-TS]Computer Science [cs]/Signal and Image Processing [STAT.AP] Statistics [stat]/Applications [stat.AP] 0103 physical sciences Classifier (linguistics) 010302 applied physics [STAT.AP]Statistics [stat]/Applications [stat.AP] Degree (graph theory) [INFO.INFO-NI] Computer Science [cs]/Networking and Internet Architecture [cs.NI] business.industry Node (networking) Aggregate (data warehouse) [MATH.MATH-IT]Mathematics [math]/Information Theory [math.IT] Pattern recognition 021001 nanoscience & nanotechnology [INFO.INFO-IT]Computer Science [cs]/Information Theory [cs.IT] Key (cryptography) Benchmark (computing) Artificial intelligence Noise (video) [INFO.INFO-IT] Computer Science [cs]/Information Theory [cs.IT] 0210 nano-technology business MathematicsofComputing_DISCRETEMATHEMATICS |
| Zdroj: | IEEE Statistical Signal Processing Workshop (SSP) IEEE Statistical Signal Processing Workshop (SSP), 2021, Rio de Janeiro (virtual), Brazil SSP |
| Popis: | International audience; Graph neural networks (GNN) have been recognized as powerful tools for learning representations in graph structured data. The key idea is to propagate and aggregate information along edges of the given graph. However, little work has been done to analyze the effect of noise on their performance. By conducting a number of simulations, we show that GNN are very sensitive to the graph noise. We propose a graphassisted Bayesian node classifier which takes into account the degree of impurity of the graph, and show that it consistently outperforms GNN based classifiers on benchmark datasets, particularly when the degree of impurity is moderate to high. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|