My primary research interest lies in the application of mathematics to develop AI algorithms that perform consistently across varied data distributions. My focus areas within deep learning include:
- Representation Learning
- Semi-supervised, self-supervised, and unsupervised methods
- General purpose models supporting multilinguality and multimodality
- Graph Neural Networks (GNNs)
- GNN applications in knowledge graphs, drug discovery, social networks, reasoning, and computational biology
- Impact of model scaling (e.g foundational models) on downstream tasks
- Intersections of machine learning and neuroscience
- Interpretation of black box models