Research groupActive
Representation Learning Group
Representation learning in graphs, knowledge graphs and weight-space models.
Representation Learning Group develops GNNs, self-supervised graph learning, knowledge graphs, LLM-graph integration, hallucination detection, LLM analysis and models that treat neural network weights as a data modality.
graph neural networksself-supervised graph learninggraph representation learningknowledge graphsmulti-graph reasoningLLMs with knowledge graphsgraph constructiongraph agentsLLM hallucination detectioninformation theoryweight-space modelsOOD detectionmodel interpretabilityanomaly detectionAI4Science
Members
· 7 people
Active Projects
No funded projects on file.
Recent Publications
- EACL 2026 Main Conference · 2026FactSelfCheck: Fact-Level Black-Box Hallucination Detection for LLMs