Generative Neural Operators through Diffusion Last Layer
Authors:
- Sungwon ParkSPSungwon Park
- Anthony ZhouAZAnthony Zhou
- Hongjoong KimHKHongjoong Kim
- Amir Barati FarimaniAFAmir Barati Farimani
Preprint
Abstract
Neural operators have emerged as a powerful paradigm for learning discretization-invariant function-to-function mappings in scientific computing. However, many practical systems are inherently stochastic, making principled uncertainty quantification essential for reliable deployment. To address this, we introduce a simple add-on, the diffusion last layer (DLL), a lightweight probabilistic head that can be attached to arbitrary neural operator backbones to model predictive uncertainty. Motivated by the relative smoothness and low-dimensional structure often exhibited by PDE solution distributions, DLL parameterizes the conditional output distribution directly in function space through a low-rank Karhunen-Loève expansion, enabling efficient and expressive uncertainty modeling. Across stochastic PDE operator learning benchmarks, DLL improves generalization and uncertainty-aware prediction. Moreover, even in deterministic long-horizon rollout settings, DLL enhances rollout stability and provides meaningful estimates of epistemic uncertainty for backbone neural operators.Keywords
All Related Versions
This publication has 14 references indexed in Scilit:
- Probabilistic weather forecasting with machine learningNature, 2024
- Neural operators for accelerating scientific simulations and designNature Reviews Physics, 2024
- A physics-informed diffusion model for high-fidelity flow field reconstructionJournal of Computational Physics, 2023
- Uncertainty quantification in scientific machine learning: Methods, metrics, and comparisonsJournal of Computational Physics, 2023
- Geometry-Informed Neural Operator for Large-Scale 3D PDEs2023
- PDE-Refiner: Achieving Accurate Long Rollouts with Neural PDE Solvers2023
- Score-based Data Assimilation2023
- B-DeepONet: An enhanced Bayesian DeepONet for solving noisy parametric PDEs using accelerated replica exchange SGLDJournal of Computational Physics, 2022
- Neural Stochastic PDEs: Resolution-Invariant Learning of Continuous Spatiotemporal Dynamics2022
- Learning nonlinear operators via DeepONet based on the universal approximation theorem of operatorsNature Machine Intelligence, 2021