Feng Chen
University of Texas at Dallas, TX·Computer Science
Artificial intelligence (AI) now helps make decisions in driving, lending, hiring, health care and security. That raises two questions: can these systems be fair, and can they know when they are unsure?
- Fields
Computer Science
Mathematics & Statistics
Engineering
- Topics
machine learning
deep learning
artificial intelligence
- Grants
- 4 known grants, through Sep 2028
What they work on
- 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.
- Apply fairness-aware online models to crime event recognition from video and purchase behavior prediction from clickstream data, examining trade-offs between accuracy and fairness.
- Design sparse and localized graph convolutional networks, neural networks for network-shaped data, that detect anomalous nodes in road, mobile, public health and geospatial networks.
- Build an active learning framework that learns which few nodes to label, reducing training labels needed for graph anomaly detection.
- Quantify multiple types of uncertainty with belief theory inside deep learning models, and test their effect on decision-making under adversarial attacks using simulations and visualization tools.
- Create training programs and hands-on projects linking AI ethics, ethical AI design and high-performance computing, including hardware acceleration for machine learning models.
More about the research
Real data streams shift over time, contain rare odd events, and come with missing or conflicting information. This research builds deep learning methods, meaning many-layered neural networks, that stay fair, flag anomalies and track uncertainty in such messy settings. It also trains researchers to build ethical AI that runs efficiently on high-performance computers.
Why it matters
Biased or overconfident AI can cause unfair loan, hiring or criminal justice decisions, or miss a person in front of a self-driving car or a cancer on a scan. These methods could make AI fairer as data change, better at spotting rare anomalies in road, mobile and public health networks, and more aware of uncertainty under attack. Training could help engineers build ethical AI that still runs fast.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Collaborative Research: CyberTraining: Implementation: Medium: Interactive and Integrated Training for Building High-Performance Ethical AI
NSF · Oct 2024 – Sep 2028
III: Medium: Collaborative Research: MUDL: Multidimensional Uncertainty-Aware Deep Learning Framework
NSF · Oct 2021 – Sep 2027
FAI: A novel paradigm for fairness-aware deep learning models on data streams
NSF · Sep 2022 – Aug 2027
Closest labs
Dong Hyun Jeong
University of the District of Columbia, DC·Computer Science
Extract time-, location- and context-aware features of network activity to predict event frequencies and estimate cyberattack severity.
Hang Liu
Rutgers University-New Brunswick, NJ·Computer Science·Early-career
Design new ways to construct and update transition probabilities for Monte Carlo graph sampling and random walk methods so sampling on GPUs runs faster.
Georgios Giannakis
University of Minnesota, MN·Engineering
Model the resonant-beam optical wireless channel, an infrared beam formed in an open-air optical cavity, using nonlinear time-varying Volterra series and kernel-based learning for both downlink and uplink.
The school
University of Texas at Dallas: 16 labs on PI Pages, $182M a year on research, 239 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.