Overview

The Meeting on Applied Algebraic Geometry (MAAG) is a regional gathering that attracts participants primarily from the Southeastern United States. MAAG 2026 takes place at Clemson University on April 18-19, 2026. Previous meetings were held at Georgia Tech in 2015, 2018, 2019, and 2023; at Clemson in 2016; and at Auburn in 2025.

Organizers: Michael Burr, John Cobb, Kisun Lee, Anton Leykin, Julia Lindberg, Luke Oeding

For any inquiries regarding the meeting, please contact [email protected].

Talks

• Yulia Alexandr (University of California, Los Angeles)
  Algebraic invariants in machine learning
  Abstract

  This talk explores how algebraic geometry offers powerful tools for understanding machine learning models. I will explain how the algebraic and semi-algebraic structure of modern models (ReLU, polynomial, and self-attention) gives rise to polynomial invariants and complexity measures, such as Euclidean distance degree. These tools provide theoretical guarantees for neural network verification, optimization, and robustness, supporting safety-critical applications.

• Ayah Almousa (University of Kentucky) — Talk Title: TBD
  Abstract

  TBD.

• Julianne Barnhart (Clemson University) — Talk Title: TBD
  Abstract

  TBD.

• Alex Dunbar (Georgia Tech) — Talk Title: TBD
  Abstract

  TBD.

• Hoon Hong (North Carolina State University) — Talk Title: TBD
  Abstract

  TBD.

• Emily King (Colorado State University)
  Group actions in frame theory, quantum information theory, and machine learning
  Abstract

  Group actions may be leveraged to create subspace configurations which are optimal in frame theory and quantum information theory, with equiangular Fourier frames and symmetric, informationally complete, positive operator-valued measures (SIC-POVMs) being two of the most well-known examples. In machine learning, highly symmetric vectors emerge when training neural networks under certain regimes (neural collapse), while orbits of group actions may be leveraged to perform classification tasks (e.g., bilipschitz invariants, group-invariant max filtering, invariant/equivariant neural networks). This talk will introduce these applications and conclude with some open problems.

• Christopher Manon (University of Kentucky) Talk Title: TBD
  Abstract

  TBD.

• Kalina Mincheva (Tulane University)
  Classifying valuated term (pre)orders
  Abstract

  Valuated term orders are studied for the purposes of Gröbner theory over fields with valuation. The points of a usual tropical variety correspond to certain valuated term preorders. Generalizing both of these, the set of all "well-behaved" valuated term preorders is canonically in bijection with the points of a "tropical adic" space. We interpret these points geometrically by explicitly characterizing them in terms of classical polyhedral geometry. This characterization gives a bijection with equivalence classes of flags of polyhedra as well as a bijection with a class of prime filters on a lattice of polyhedral sets.

Schedule

All talks in Martin Hall.

Participant Information

Funding

This conference is supported all or in part by the National Science Foundation under DMS Award No. 2552457.

To be considered for travel support, please register by March 15, 2026. Students and early career mathematicians are strongly encouraged to present a poster; priority for travel funding will be given to poster presenters.