Graph Machine Learning and Artificial Intelligence: Aref Einizade's Research on Complex Graphs
Aref Einizade is an expert in artificial intelligence, specifically in signal processing and machine learning. It was during his studies at Sharif University of Technology in Tehran that the young researcher decided to pursue this path. During his PhD, he focused in particular on the processing of medical and biomedical data. “I was interested in brain signals and how to represent the connectivity between different regions of the brain in patients with disease or undergoing sleep testing. I based my work on graph machine learning and graph signal processing,” explains Aref Einizade.
Managing large-scale interconnected data
This mathematical and computer science discipline makes it possible to effectively represent data relating to objects interacting in a network. Each object is then a node, and the nodes are connected to each other by edges (links). “In my doctoral work, for example, the different regions of the brain were considered as nodes, allowing me to study the connectivity between each of them,” explains the researcher.
In November 2023, Aref Einizade began a postdoctoral fellowship at the Information Processing and Communication Laboratory (LTCI***) at Télécom Paris. He joined the team of Jhony Giraldo, senior lecturer, and embarked on more theoretical and experimental work on graph topology—namely, the study of their overall structure, the dynamics of the information propagation across the graph topology, the connectivity between different objects, etc.
The researcher then focused on understanding and modeling complex multi-domain graphs. This was a real challenge, as it involved processing large-scale interconnected data sets. Topological Deep Learning—which combines Deep Learning and graph topology—made it possible to consider not only the values of the data but also their overall structure. “It offers a flexibility that has allowed me to develop algorithms capable of handling a variety of graphs and reaching a new frontier in machine learning,” says Aref Einizade enthusiastically.
Better understanding brain activity
This work, supported by the Hi! Paris interdisciplinary center and ANR projects DeSNAP, also aims to identify the various mathematical problems involved and provide theoretical guarantees. The researcher is exploring several possible applications, including weather forecasts, which require analyzing data from networks of observation stations, and social networks and their various communities.
In terms of biomedical applications, Aref Einizade naturally made the connection with his doctoral work. “The power of graphs associated with deep learning reveals the evolution of brain activity over time. It becomes possible to map the connections that are created between regions of the brain and their dynamics.” The researcher then applied his algorithms to databases from electroencephalograms (EEGs) of patients monitored during sleep, tasks, or suffering from cognitive diseases. “The results obtained are consistent with the neuroscientific literature. This is particularly encouraging. In my opinion, graphs have great potential for detecting the origin of many problems and leading to their resolution.”
Now holding a tenure track research contract supported by IP Paris's interdisciplinary Engineering 4 Health (E4H) center, Aref Einizade is continuing his work to develop more robust and scalable machine learning solutions based on graphs, with applications in neuroscience and biomedicine in particular. “Thanks to the Tenure Track, I have access to the particularly stimulating environment of the Saclay plateau,” says the researcher. “I rub shoulders with people who are inspiring, both scientifically and in their lifestyle, and that's very motivating.”
*As part of the DaTSHealth project selected by the ANR during the call for projects “Skills and professions of the future” (AMI CMA) France 2030
** AMI-CMA DaTSHealth : ANR-23-CMAS-0033
About Aref Einizade
Aref Einizade has been a postdoctoral researcher (Nov. 2023–Jan. 2026) in the Multi-Media team at Télécom Paris, Institut Polytechnique de Paris. His research lies at the intersection of graph machine learning, graph signal processing, and graph neural networks, with a focus on theoretically grounded methods for multi-modal graphs and applications in biomedical data (especially brain signals), weather, and traffic forecasting. He received his Ph.D. in Electrical Engineering from Sharif University of Technology in February 2023, where he developed GSP and GML algorithms for unknown graph structures, primarily for biomedical applications, such as modeling the connections between different brain regions using graph analytics. His recent work extends to generalized structures, such as simplicial complexes and hypergraphs, with potential future applications in biomedical signals and images. His work has been published in leading international conferences (e.g., NeurIPS, ECML, and IEEE SPMB) and peer-reviewed top-tier journals, including Neural Networks (Elsevier), IEEE Transactions on Signal and Information Processing over Networks, IEEE Transactions on Cognitive and Developmental Systems, IEEE Signal Processing Letters, Biomedical Signal Processing and Control, Neuroscience Informatics, and Digital Signal Processing.
***LTCI: a research lab Télécom Paris, Institut Polytechnique de Paris, 91120 Palaiseau, France