Developing cutting-edge algorithms at the intersection of machine learning and mathematical optimization to solve large-scale decision-making problems.
Click on a topic to explore related publications and projects
Leveraging generative models and deep learning to solve complex combinatorial optimization problems more efficiently.
Scaling optimization algorithms to distributed multi-GPU systems for industrial-scale problem instances.
Using machine learning to learn powerful heuristics for combinatorial optimization problems
The first generative framework that models the joint distribution of both integer and continuous variables for MILP solutions. Achieves 41.34% improvement over existing baselines with holistic guidance mechanism.
Scaling primal-dual algorithms to distributed multi-GPU architectures
The first distributed PDHG-based solver for large-scale LPs that scales across multiple GPUs. Achieves near-linear speedup with up to 6× faster solving on 8 GPUs while maintaining full FP64 numerical accuracy.