Hang Liu
Rutgers University-New Brunswick, NJ·Computer Science·Early-career
Modern computing faces two linked challenges: handling enormous datasets fast, and making artificial intelligence (AI) both fast and fair. Graphs, networks of nodes linked by edges, describe pandemic spread, chip designs and software flaws.
- Fields
Computer Science
Engineering
- Topics
graph sampling
random walk
graph analytics
- Grants
- 2 known grants, through Sep 2028
What they work on
- Design new ways to construct and update transition probabilities for Monte Carlo graph sampling and random walk methods so sampling on GPUs runs faster.
- Recast core graph sampling steps to exploit GPU tensor cores and ray-tracing cores, and use RDMA-assisted adaptive scheduling to scale to trillion-edge graphs.
- Build a bias-centric programming framework and open software library that lets users express existing and future graph sampling algorithms while hiding low-level optimizations.
- Develop nine training modules spanning philosophical AI ethics, ethical AI and HPC hardware and software, with hands-on projects such as hardware acceleration for machine learning models.
More about the research
Some now have trillions of edges, too many to analyze directly. Meanwhile, AI trained on biased data can produce unfair outcomes in hiring, lending or criminal justice. The tricks used to speed up AI can also make it less fair. The lab's two projects address these problems from the system and the training sides.
Why it matters
A fast, user-friendly GPU library for trillion-edge graphs could help scientists and engineers in academia, national laboratories and industry. It could support work such as modeling pandemic spread, designing large integrated circuits and finding software vulnerabilities. Training in ethical, high-performance AI could help developers build systems that are fast without producing unfair outcomes.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Feng Chen
University of Texas at Dallas, TX·Computer Science
Develop online fairness-aware deep learning models for changing data streams, using causal representation learning, differentiable architecture search and online meta-learning, and release open software.
Shivaram Venkataraman
University of Wisconsin-Madison, WI·Computer Science·Early-career
Characterize performance variability across many types of accelerator hardware, where identical applications run at different speeds on chips of the same type, and identify which applications are most affected (Fortuna).
Peng Jiang
University of Iowa, IA·Computer Science·Early-career
Develop a runtime system that gives a virtualized view of data partitioned across heterogeneous devices, plus a compiler tool that places data using a statistical performance model of random memory accesses.
The school
Rutgers University-New Brunswick: 57 labs on PI Pages, $837M a year on research, 713 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.