Learning data-driven structure and dynamics for lab-in-the-loop AI scientists

THE SPEAKER

Smita Krishnaswamy, PhD - Professor, Yale University

Smita Krishnaswamy is a Professor of Computer Science and Genetics at Yale University. She is also affiliated with the Program for Applied Mathematics and WTI Institute for NeuroComputation and Machine Intelligence at Yale and an affiliate member of the MILA Quebec AI Institute. Smita's lab works on fundamental deep learning and mathematical machine learning methods for accelerating discovery from biomedical and neuroscientific data. Her work features many methods for generative modeling, graph-based learning, visualization, dynamics modeling and optimal transport as well as multimodal integration of high dimensional data. She has applied her techniques to discovery from cellular, molecular, and imaging data from neuroscience, psychology, stem cell biology, cancer, and immunology. Smita obtained her Ph.D. from the University of Michigan in Computer Science. Smita's work has won several awards including the NSF CAREER Award, Sloan Faculty Fellowship, and Blavatnik Fund for Innovation. Smita teaches deep learning, unsupervised learning, geometry topology in ML and other courses at the intersection of CS and applied math. She also teaches special courses in computational genomics at CGSI (UCLA), CSHL, as well as being a mentor for the Yale SUMRY math REU program.

WEBINAR SUMMARY

In this talk, Dr. Krishnaswamy will cover progress towards building a fundamentally new type of AI scientist that goes beyond orchestrating existing analysis pipelines or selecting among pre-defined tools. Instead, this scientist infers mechanistically plausible generative models of complex, systems-level data by unifying mathematical modeling with deep learning. Unlike current AI scientists, which primarily automate experimental design, literature synthesis, or statistical analysis, our approach seeks to learn an explicit, interpretable model of the underlying system itself. For example, in cellular data, the system infers an abstract but mechanistically grounded model of a cell that could generate the observed molecular measurements. Towards this end I will cover "ingredients" of this that we have been developing including geometric data representations, data shape detection, flow-based models for trajectory inference, and graph and sheaf ODE models for dynamics and perturbation modeling. These tools can then be orchestrated by an LLM agent.

 
Learning Objectives:

Participants will develop an understanding of:
- Representation Learning, Trajectory Inference, Dynamics modeling, AI Scientists, Lab-in-the-loop AI

Who Should Attend:
Computational Biologists/Bioinformaticians, Data Analysts, Educators/Trainers, Industry Scientists (vendor-agnostic, tool developers, method innovators) Research Scientists, Share Resource Laboratory (SRL) Staff,  Trainees (graduate students, postdocs, early-career researchers), Translational Immunologists

Keywords: Representation Learning, Trajectory Inference, Dynamics modeling, AI Scientists, Lab-in-the-loop AI

CMLE Credit: 1.0