Find a story
Search Spins
Search titles, summaries, and missing voices across published articles — press releases, announcements, and media coverage.
5 results for “neural operator”
Disentangling Attention in Deep Operator Learning: A Controlled Study of Data-Driven and Physics-Informed Architectures
A controlled study isolates how specific attention mechanisms affect accuracy in deep neural operators for PDE solving, finding cross-attention consistently improves performance while self-attention’s benefits depend on problem complexity.
Sep 7, 2026
L-FNO: Lorentzian Fourier Neural Operator for Stochastic Event Dynamics
Researchers introduced L-FNO, a new stochastic neural operator architecture designed to model rare, bursty, self-exciting events in operational systems by integrating Lorentzian spectral kernels and likelihood-based training — improving event likelihood estimation, calibration, and rare-event detection over existing neural operator baselines.
Aug 17, 2026
Geometry-aware Incremental Neural Operator for Long-Horizon PDE prediction
Researchers introduced GeoIncNO, a new neural operator architecture designed to improve stability and accuracy in long-horizon autoregressive prediction of partial differential equations by structuring latent increments with geometric awareness and decoupling mean and fluctuation reconstruction.
Aug 13, 2026
Feature Interaction Modeling for Physics-Informed Neural Networks and Neural Operators
Researchers introduced feature interaction modules from factorization machines into physics-informed neural networks and neural operators to improve accuracy on parameterized PDEs with strong cross-variable dependencies, especially shock-dominated or discontinuous systems.
Aug 3, 2026
Diffusion-corrected Autoregressive Fourier Neural Operator for Droplet Evolution Prediction
Researchers introduced DiffARFNO, a two-stage neural operator combining autoregressive Fourier-MIONet with a DDIM-based corrector to improve long-horizon droplet evolution prediction in inkjet printing simulations.
Jul 21, 2026