Oak Ridge National Laboratory — Knoxville, TN
Distinguished R&D Staff · Group Leader, Discrete Algorithms
Accelerating scientific discovery through distributed machine learning and graph algorithms on HPC platforms — reducing computation from weeks to seconds.
SymProp: Scaling Sparse Symmetric Tucker Decomposition via Symmetry Propagation
AI-Powered Knowledge Graphs for Neuromorphic and Energy-Efficient Computing
Accelerated Constrained Sparse Tensor Factorization on Massively Parallel Architectures
Exaflops Biomedical Knowledge Graph Analytics
PLANC: Parallel Low-rank Approximation with Nonnegativity Constraints
Distributed ML and graph algorithms on HPC platforms, accelerating scientific data analysis at exascale.
Integrating domain-specific scientific knowledge and physical constraints with machine learning models.
Matrix and tensor factorization, nonnegative decompositions, and low-rank approximation at scale.
All-pairs shortest paths, graph analytics, and sparse linear algebra on multi-GPU and distributed systems.
DOE Advanced Scientific Computing Research highlights research on machine learning for scientific data.
Team demonstrated 1 ExaFLOP performance on a biomedical knowledge graph application, recognized with the UT-Battelle Research Accomplishment Award.
Led SNAPSHOT project, ACM Gordon Bell Prize finalist, developing AI methods to mine COVID-19 connections on Summit supercomputer.
Science-guided ML research on Summit leads to cleaner aviation fuel efficiency gains.
Co-authored Book — 2022
A comprehensive treatment of integrating scientific domain knowledge with machine learning for improved accuracy, interpretability, and generalizability.
Learn more →