Differences

This shows you the differences between two versions of the page.

--- aira:start [2026/05/21 14:46] – [Schedule Spring 2026] mtm
+++ aira:start [2026/05/28 14:30] (current) – [Schedule Spring 2026] mtm
@@ Line 21: / Line 21: @@
 ===== Schedule Spring 2026 =====
+  * **[PHD TRACK] 2026.05.28**: Soheila Molaei,  senior researcher @ University of Oxford, [[#section20260528|From Graphs to Graph Neural Networks: Foundations and Applications in Healthcare]]
+    * Meeting link:[[https://teams.microsoft.com/meet/377493387003987?p=TVBcvsqGs42CdT7U9T|MS Teams]]
+    * Recording:  [[https://ujchmura.sharepoint.com/:v:/t/Section_495645_1/IQC8c4XxDZvTQ70h5GCoIrRwAYBNQyagBm3DaEdsBovDMNQ?e=8fwmHj|View]]
+    * Presentation slides:  TDA
   * **[RESEARCH TRACK] 2026.05.21**: Jan Argasiński,  assistant professor @ Jagiellonian University, [[#section20260521|Computational Neuroscience at Sano (Centre for Computational Medicine): Current Research and a Spotlight on “A Tract Density Biomarker for Survival Prediction in Glioblastoma”]]
@@ Line 565: / Line 570: @@
+==== 2026-05-28 ====
+<WRAP column 15%>
+{{ :aira:soheila_molaei_foto.png?200| }}
+</WRAP>
+<WRAP column 75%>
+**Speaker**: Soheila Molaei,  Senior researcher @ University of Oxford
+**Title**: From Graphs to Graph Neural Networks: Foundations and Applications in Healthcare
+**Abstract**:
+In this talk, I will introduce the foundations of graph theory and graph neural networks, starting from intuitive examples of real-world graphs such as social networks, molecules, road networks, and biomedical interaction networks. I will explain why graph-structured data challenges standard machine learning assumptions, and how GNNs use message passing and representation learning to model complex relationships. The talk will then cover common GNN tasks, including node classification, link prediction, and graph classification, before introducing key models such as graph convolutional networks and graph attention networks. I will conclude with applications in healthcare, including medical graphs, drug discovery, and patient modelling.
+**Biogram**:
+Soheila Molaei is a senior researcher in the Department of Engineering Science at the University of Oxford. Her research focuses on artificial intelligence and machine learning, particularly graph neural networks, federated learning, neuro-symbolic AI, and learning from multimodal and heterogeneous data, with applications in complex real-world and healthcare domains.
+</WRAP>
+<WRAP clear></WRAP>
 ==== 2026-05-21 ====
@@ Line 577: / Line 601: @@
 **Title**: Computational Neuroscience at Sano (Centre for Computational Medicine): Current Research and a Spotlight on “A Tract Density Biomarker for Survival Prediction in Glioblastoma”
-**Abstract**:
+Abstract:
-Significant computational capabilities of modern FPGAs, combined with high-level methodologies for developing their configuration, open new areas where unique features of this technology can be exploited, such as real-time processing, deterministic latency, reconfigurability, and low power consumption. As part of my research, I decided to explore the capabilities of this technology across data with various characteristics, develop a set of best-practice data processing techniques, and verify them using real-world use cases originating from large scale physics-experiments, medical imaging and computer networks.
+Computational neuroscience provides powerful tools to better understand brain structure and function, and to translate this knowledge into clinically relevant biomarkers for neurological disease. In this seminar, I will briefly survey ongoing projects in brain modelling, advanced neuroimaging analysis, and machine learning for neurological and psychiatric disorders, with a particular emphasis on how these methods bridge basic science and clinical practice. The second part of the talk will spotlight the development of a tract density–based biomarker aimed at predicting survival in patients with glioblastoma, illustrating the full pipeline from diffusion MRI processing and tractography through feature extraction to predictive modelling and validation. By showcasing both the broader research landscape at Sano and this focused case study, the seminar will highlight how computational approaches can support prognosis, treatment planning, and ultimately more personalized care in neuro-oncology.
-**Biogram**:
+Biogram:
-Expert in the field of Field Programmable Gate Arrays (FPGA) technology with many years of experience acquired while working in international research projects. Ph.D. in technical sciences in the discipline of computer science obtained for the design and implementation of the data acquisition system for the HADES experiment detector system, which has also been used in dozens of other applications. Popularizer of FPGA technology by organizing conferences and training program in this field on a national scale. Since 2018 conducts research on the use of FPGAs in subjects related to processing massive amount of streamlined data such as in High Performance Computing, low and fixed latency networking. Technical coordinator of the Data Acquisition System in the PANDA experiment.
+Jan Argasiński, PhD, is a researcher at the Department of Human-Centered Artificial Intelligence, Faculty of Physics, Astronomy and Applied Computer Science, Jagiellonian University in Kraków. He also leads the Computational Neuroscience Group at Sano – Centre for Computational Medicine. His research focuses on computational neuroscience, neuromorphic computing, affective computing, serious games, and VR/AR.
 </WRAP>
 <WRAP clear></WRAP>