Research

Deep Learning for Knot Classification

• Using various deep learning architectures to accurately classify different low-dimensional topologies (specifically knots) and to learn novel relationships between geometry and topology.

Deep Learning for Knot Invariant Relationships

• Probing conjectured relationships between invariants of knots using data analysis and machine learning methods.

Mechanistic Interpretability of Protein Language Models

• Sifting through the latent space of pLMs to understand how they work in the context of geometry and folding.

Knot Theoretic Measures for Protein Family Classification

• Investigating the relationship between hierarchical protein families (CATH) and structural features motivated from geometry and topology.