Xu Yuan
University of Delaware, DE·Computer Science·Early-career
Huge, messy, constantly changing data streams, from satellite images of the ocean to posts on social media, hide patterns that people cannot find by hand. This research builds artificial intelligence (AI) methods, shared datasets, and software toolkits to extract that structure.
- Fields
Computer Science
Earth & Environmental Science
Engineering
- Topics
artificial intelligence
machine learning
foundation models
- Grants
- 2 known grants, through Sep 2028
What they work on
- Construct an open, large-scale database of estuarine and coastal phytoplankton observations to address the scarcity of field data, with global sharing and contribution.
- Build the first foundation model for phytoplankton community composition from hyperspectral satellite imagery, and use AI to predict changes in plankton communities.
- Develop AI methods that capture spatial, spectral, and temporal variation in coastal phytoplankton communities, and release them as open software toolkits.
- Develop a real-time detector for malicious social network content that copes with shifting features, scalable processing, and scarce labeled data.
- Design a self-monitoring, evolving malicious knowledge graph linking content, concepts, and actors, and use periodic analysis to reveal stealthy malicious communities, evaluated on real social networks.
More about the research
One project tracks microscopic ocean plants. The other detects malicious behavior online.
Why it matters
Satellite-based AI tools for reading phytoplankton communities could give oceanographers a regional-to-global view of shifts in coastal food webs and fisheries production under a changing climate. Real-time detection and modeling of malicious online content could help social media platforms reduce spam, deception, and political manipulation, along with the economic losses they cause.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Bharath Hariharan
Cornell University, NY·Computer Science·Early-career
Build recognition systems that learn difficult visual concepts from as few as 1–2 labeled images and about 1,000 unlabeled ones, by drawing on a memory of past visual tasks across domains such as microscopy.
Carl Vondrick
Columbia University, NY·Computer Science·Early-career
Build open-world recognition models for satellite, aerial and ground imagery from vision-language models without language annotations, using cross-modal contrastive learning and reasoning about change over time.
Adnan Rajib
University of Texas at Arlington, TX·Earth & Environmental Science
Evaluate the scalability and cost-effectiveness of the Living Engineered Reef System on the Texas Gulf Coast through a pilot study, techno-economic assessment (weighing performance against cost), and community co-design workshops.
The school
University of Delaware: 22 labs on PI Pages, $466M a year on research, 298 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.