Philip Yu
University of Illinois Chicago, IL·Computer Science
Modern data are vast, and two practical problems run through this work: spotting the unusual and suggesting the relevant. Anomaly detection finds data points that deviate from the norm, which can signal security breaches, health crises or financial fraud.
- Fields
Computer Science
Mathematics & Statistics
- Topics
anomaly detection
outlier detection
open-source software
- Grants
- 2 known grants, through May 2027
What they work on
- Integrate open-source anomaly detection tools for tabular, graph and time-series data into a single OpenAD ecosystem with a standardized development environment for building and evaluating models.
- Develop OpenAD tools that automate and accelerate anomaly detection, and establish a governance structure to guide community contributions to the platform.
- Build a self-supervised recommender framework on a graph neural network pre-trained with data augmentation and contrastive learning, then fine-tune it for fair recommendations for minority user groups.
- Address cold-start users with little data, cross-platform learning, and changing user behavior through data augmentation, inter-platform contrastive learning and lifelong self-supervised model tuning.
More about the research
Recommender systems suggest items, treatments or contacts that users are likely to want. Both are machine-learning problems, and both suffer when tools are fragmented or labeled data are scarce.
Why it matters
Easier-to-use anomaly detection tools could help hospitals, security agencies, banks and businesses catch fraud, breaches or health crises sooner. Recommender systems that need fewer labels and treat minority user groups fairly could improve services on online platforms. Suggested uses include e-commerce, social networks, e-learning and recommending targeted treatments in precision medicine.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Meng Jiang
University of Notre Dame, IN·Computer Science·Early-career
Develop and compare natural language generation models, one data-driven and one based on theories of abduction, for comparative summarization and argument generation in scientific domains.
Yufei Huang
University of Pittsburgh, PA·Computer Science
Develop a contrastive learning-based geometric graph attention model that learns gene expression patterns aware of tissue shape from single-cell spatial transcriptomics data, so that the patterns match the tissue structure seen in histology.
Jurij Leskovec
Stanford University, CA·Computer Science
Develop multi-level graph neural networks that handle graphs whose nodes are themselves complex structures, such as molecular graphs, for scientific foundation models.
The school
University of Illinois Chicago: 42 labs on PI Pages, $532M a year on research, 359 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.