ICLR 2026 Orals

Quotient-Space Diffusion Models

Yixian Xu, Yusong Wang, Shengjie Luo, Kaiyuan Gao, Tianyu He, Di He, Chang Liu

Diffusion & Flow Matching Sat, Apr 25 · 3:27 PM–3:37 PM · 201 C Avg rating: 7.50 (6–10)
Author-provided TL;DR

We propose a principled way to leverage group symmetry of the target distribution by defining a diffusion model on the quotient space, which achieves both easier learning and correct sampling for the first time.

Abstract

Diffusion-based generative models have reformed generative AI, and have enabled new capabilities in the science domain, for example, generating 3D structures of molecules. Due to the intrinsic problem structure of certain tasks, there is often a symmetry in the system, which identifies objects that can be converted by a group action as equivalent, hence the target distribution is essentially defined on the quotient space with respect to the group. In this work, we establish a formal framework for diffusion modeling on a general quotient space, and apply it to molecular structure generation which follows the special Euclidean group SE(3) symmetry. The framework reduces the necessity of learning the component corresponding to the group action, hence simplifies learning difficulty over conventional group-equivariant diffusion models, and the sampler guarantees recovering the target distribution, while heuristic alignment strategies lack proper samplers. The arguments are empirically validated on structure generation for small molecules and proteins, indicating that the principled quotient-space diffusion model provides a new framework that outperforms previous symmetry treatments.

One-sentence summary·Auto-generated by claude-haiku-4-5-20251001(?)

Quotient-space diffusion models reduce learning difficulty for molecular structure generation via SE(3) symmetry handling.

Contributions·Auto-generated by claude-haiku-4-5-20251001(?)
  • Establishes formal framework for diffusion modeling on quotient spaces with respect to group actions
  • Reduces necessity of learning group action component, simplifying learning difficulty
  • Guarantees sampler recovers target distribution while removing unnecessary group-direction movement
  • Demonstrates better generation quality and design success rate on small molecules and proteins
Methods used·Auto-generated by claude-haiku-4-5-20251001(?)
  • Diffusion models
  • Quotient space mathematics
  • SE(3) symmetry
  • Group-equivariant methods
Datasets used·Auto-generated by claude-haiku-4-5-20251001(?)
  • GEOM-QM9
  • GEOM-DRUGS
Limitations (author-stated)·Auto-generated by claude-haiku-4-5-20251001(?)

Authors did not state explicit limitations.

Future work (author-stated)·Auto-generated by claude-haiku-4-5-20251001(?)

Authors did not state explicit future directions.

Author keywords

  • Diffusion Models
  • Generative Modeling
  • Geometric Deep Learning
  • Molecular Structure Generation

Related orals

Generative Human Geometry Distribution

Introduces distribution-over-distribution model combining geometry distributions with two-stage flow matching for human 3D generation.

Avg rating: 5.50 (2–8) · Xiangjun Tang et al.
Something off? Let us know →