Xiaoyu Huang
Temple University, PA·Mathematics & Statistics
Mathematicians increasingly work with huge collections of objects, such as numbers, curves and functions, where subtle patterns are hard to see by hand. Artificial intelligence (AI) can spot patterns, but its findings are often opaque and unproven.
- Fields
Mathematics & Statistics
Computer Science
- Topics
number theory
combinatorics
AI for mathematics
- Grants
- 1 known grant, through Aug 2029
What they work on
- Generate and curate large public datasets of number-theoretic and combinatorial objects, including a database of noncongruence modular forms for studying their arithmetic properties.
- Apply machine learning to detect patterns, then use interpretability tools that expose a model's internal logic to turn those patterns into precise conjectures and algorithms.
- Validate AI-derived claims by rigorous proof, automated theorem provers or AI-guided searches for counterexamples.
- Use reinforcement learning to optimize equations for algebraic curves and to train agents that discover explicit mathematical mappings for unsolved conjectures.
More about the research
This collaborative project aims to build a rigorous way for mathematicians and AI to work together on core problems in number theory and combinatorics, the mathematics of counting and arrangements. Patterns found by AI would be turned into clear, proven mathematical statements.
Why it matters
The work could produce new results in number theory and combinatorics. It could also offer a reproducible way to use AI for discovery that stays interpretable and verifiable. Other data-heavy sciences could adopt these pattern-finding methods. Public datasets, software and teaching materials, including a database of noncongruence modular forms, would support future researchers.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Jesus De Loera
University of California-Davis, CA·Mathematics & Statistics
Study slices and shadows of polytopes, meaning their cross-sections and projections, using enumerative and extremal combinatorics with computer-aided experimentation.
Ken Ono
University of Virginia, VA·Mathematics & Statistics
Develop algebraic, analytic and combinatorial tools for harmonic Maass forms, generalizations of modular forms that underlie Ramanujan's mock theta functions, for attacking open problems in number theory.
Jeremy Kahn
Brown University, RI·Mathematics & Statistics
Determine whether three or four definitions of a surface subgroup being "closely aligned" with an sl_2(R) subgroup, inside lattices of semi-simple Lie groups, are mathematically equivalent.
The school
Temple University: 22 labs on PI Pages, $315M a year on research, 238 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.