Representation learning on graphs enables downstream tasks such as graph classification and link prediction in multiple application scenarios. We design novel graph representation learning pipelines to achieve state-of-the-art performance on these downstream tasks, leveraging tools such as random-walks, social theories, graph neural networks, and bilevel optimization. We also investigate the explainability of graph representation learning models, which is vital, especially in high-stakes applications such as drug discovery and anomaly detection.
Past Research Projects
For more information on the lab's previous research projects, please see our publication page.