Structured Neural SDEs for Functional Calibration
Abstract
Neural Stochastic Differential Equations (Neural SDEs) provide flexible continuous-time generative models, but generic neural drift and diffusion networks are costly to simulate on long horizons and can give unstable gradients when the training signal is a path functional rather than a pointwise observation. We introduce SLiSDE, a family of Neural SDE models built from structured linear stochastic layers. Parallel-in-time simulation is obtained at the layer level, while expressivity is recovered by gated in-flow stacking: previous-layer paths modulate the next layer's latent flow through learned gates. For functional calibration tasks in which rare paths dominate the loss, we add an optional Girsanov tilt that acts as a learned importance sampler with an exact likelihood-ratio correction. We prove well-posedness, a discretisation error bound, validity of the change of measure, and a universality result: the terminal laws of the gated stack are dense in the space of square-integrable laws. Experiments on functional calibration benchmarks show that the structured model outperforms fully neural SDE baselines while retaining parallel-time simulation and stable importance weights.
Related
- Exact Controllability of Backward-Structured Mean-Field SDEs: Hautus Criteria and Initial-Time Dichotomy
- Structured SDEs for Realized GARCH: Continuous-Time Volatility Modelling and Component-Level Analysis
- The Numerical Onsager-Machlup action functional for Euler discretized SDEs driven by fractional Brownian motion
- SDECast: Probabilistic Weather Forecasting in Continuous Time with Neural SDEs
- HyperNSDE: Personalized Neural SDEs for Joint Static-Longitudinal Clinical Data Generation
BibTeX
@misc{piatti2026structured,
title = {{Structured Neural SDEs for Functional Calibration}},
author = {Francesco Piatti and Andrea Iannucci and Thomas Cass},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.34831},
eprint = {2609.34831},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.34831},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Structured Neural SDEs for Functional Calibration
AU - Francesco Piatti
AU - Andrea Iannucci
AU - Thomas Cass
PY - 2026
DA - 2026/09/28
PB - arXiv
DO - 10.48550/arxiv.2609.34831
UR - https://arxiv.org/abs/2609.34831
AB - Neural Stochastic Differential Equations (Neural SDEs) provide flexible continuous-time generative models, but generic neural drift and diffusion networks are costly to simulate on long horizons and can give unstable gradients when the training signal is a path functional rather than a pointwise observation. We introduce SLiSDE, a family of Neural SDE models built from structured linear stochastic layers. Parallel-in-time simulation is obtained at the layer level, while expressivity is recovered by gated in-flow stacking: previous-layer paths modulate the next layer's latent flow through learned gates. For functional calibration tasks in which rare paths dominate the loss, we add an optional Girsanov tilt that acts as a learned importance sampler with an exact likelihood-ratio correction. We prove well-posedness, a discretisation error bound, validity of the change of measure, and a universality result: the terminal laws of the gated stack are dense in the space of square-integrable laws. Experiments on functional calibration benchmarks show that the structured model outperforms fully neural SDE baselines while retaining parallel-time simulation and stable importance weights.
ER -