The क्रम (Krama) Lab is a research group steered by the research interests of Dr. Manohar Kaul. Our research projects primarily focus on applied algebraic topology (topological data analysis), optimal transport theory (discrete OT and assignment / matching problems), and geometric deep learning (graph and point-cloud representation learning, knowledge graph representations). The work aims to propose machine learning algorithms that are both practical and theoretically well-founded. An overarching principle underlying most of our lab's research is to produce efficient learning algorithms that exploit the geometric and/or topological information present in data, which can manifest itself as latent higher-order structures, to perform well on limited training examples.
Our work has resulted in publications in top-tier venues for machine learning (e.g., ICML, NeurIPS, ICLR), NLP (e.g., ACL) and computer vision (e.g., ECCV). For more detailed information on our publications, please visit our publications page. We are actively involved in training new students on our research topics via formal courses (e.g., CS5710 Computational Topology) and informal research brainstorming sessions.
If you are interested in becoming a part of the क्रम (Krama) Lab, review our Join page.