Our research is broadly focused on machine learning, with a particular interest in developing methods that remain effective when computational resources, training data, or the quality and completeness of observations are limited. This interest has grown from working with scientific and biomedical datasets, where data are often small, imbalanced, heterogeneous, partially observed, or naturally represented through structured relationships.

For example, FUSE develops efficient node representation learning from graph structure when node features are absent or unreliable, while MSDE detects anomalies without assuming a particular anomaly mechanism by examining how observations respond to density enhancement. In a different setting, SADP develops a computationally efficient and local learning rule for spiking neural networks, avoiding the need for precise spike-pair timing. These methods reflect our broader interest in designing learning algorithms that extract useful information from limited data while reducing computational demands.

Rather than relying primarily on increasingly large models, we aim to develop efficient, data-aware learning algorithms that make better use of the information that is actually available. Our work spans machine learning methodology, graph and hypergraph learning, anomaly detection, multimodal learning, and applications in scientific and biomedical domains.

While some of our applications are rooted in biomedical and clinical sciences, the algorithms we develop are domain-agnostic and broadly applicable to graph data, time series, and high-dimensional tabular datasets. We strongly value collaborative research, intellectual honesty, and methodological rigor, and actively seek interdisciplinary collaborations that push the boundaries of data-driven scientific discovery.

At present, we are not accepting applications for internships. Students interested in pursuing a minor project aligned with the research directions of the lab may apply. Application emails should include a CV, an academic transcript, and a brief statement of research interests clearly aligned with the lab’s focus areas. We accept major project students exclusively from the School of Data Science.