Measure Neural Mappings
Abstract
We introduce and define measure neural mappings (MNMs), which combine measure-dependent neural transport with ensemble consistency for learning operators on probability measures. Ensemble consistency requires outputs from growing ensembles sampled from an input law to converge to the specified population output in probability. We establish quantitative continuum limits for attention on noncompact laws under strong sub-Gaussian and tail-integrability assumptions. These laws accommodate unbounded state spaces in scientific machine learning, including Bayesian inference and uncertainty quantification. For fixed self-attention transformers, the expected output Wasserstein error is bounded by the empirical input Wasserstein error plus an additional N−1/2 attention term. These self-attention transformers are MNMs under the stated assumptions: pre-norm architectures follow from compact stability after normalization, while post-norm architectures are covered under the stated tail conditions. We prove corresponding cross-attention estimates and extend the results to multihead attention and grouped transformer evaluation, with convergence as the group size grows. For Set Transformers viewed as measure-to-vector mappings, we obtain an N−1/2 bound on the expected Euclidean output error under the stated assumptions. Experiments on seven operators in dimensions d=1,2,3,5,10 support the predicted particle-count dependence, using one-dimensional Wasserstein computations and Sinkhorn approximations in higher dimensions. The observed Wasserstein and sliced Wasserstein decay rates are close to their respective theoretical sampling benchmarks.
Related
- Best proximity point results of operator type condensing mappings via measure of noncompactness and application to an infinite system of differential equations in c0 and ℓ1
- Sharp Bohr and Bohr-Rogosinski inequalities involving area measure for close-to-convex harmonic mappings
- Depth Separation for Neural Networks under Gaussian Measure
- Neural Harmonic Measure Operator
- Radon Measure Representations for Infinite-Width Neural Networks with Singular Activations
BibTeX
@misc{baptista2026measure,
title = {{Measure Neural Mappings}},
author = {Ricardo Baptista and Edoardo Calvello and Bohan Chen and Andrew Stuart and Urbain Vaes},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.17384},
eprint = {2610.17384},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.17384},
}
RIS
TY - UNPB
TI - Measure Neural Mappings
AU - Ricardo Baptista
AU - Edoardo Calvello
AU - Bohan Chen
AU - Andrew Stuart
AU - Urbain Vaes
PY - 2026
DA - 2026/10/12
PB - arXiv
DO - 10.48550/arxiv.2610.17384
UR - https://arxiv.org/abs/2610.17384
AB - We introduce and define measure neural mappings (MNMs), which combine measure-dependent neural transport with ensemble consistency for learning operators on probability measures. Ensemble consistency requires outputs from growing ensembles sampled from an input law to converge to the specified population output in probability. We establish quantitative continuum limits for attention on noncompact laws under strong sub-Gaussian and tail-integrability assumptions. These laws accommodate unbounded state spaces in scientific machine learning, including Bayesian inference and uncertainty quantification. For fixed self-attention transformers, the expected output Wasserstein error is bounded by the empirical input Wasserstein error plus an additional N^−1/2 attention term. These self-attention transformers are MNMs under the stated assumptions: pre-norm architectures follow from compact stability after normalization, while post-norm architectures are covered under the stated tail conditions. We prove corresponding cross-attention estimates and extend the results to multihead attention and grouped transformer evaluation, with convergence as the group size grows. For Set Transformers viewed as measure-to-vector mappings, we obtain an N^−1/2 bound on the expected Euclidean output error under the stated assumptions. Experiments on seven operators in dimensions d=1,2,3,5,10 support the predicted particle-count dependence, using one-dimensional Wasserstein computations and Sinkhorn approximations in higher dimensions. The observed Wasserstein and sliced Wasserstein decay rates are close to their respective theoretical sampling benchmarks.
ER -