analiza aiqbench
This arXiv preprint argues that simpler, sparser surrogate models can beat exact high-dimensional simulators on control tasks, such as a modified Lunar Lander, because lower dimensionality makes learning and control easier. Its core contribution is a method that uses conditional generative modeling and Gaussian-process distillation to find which variables matter for the full conditional distribution, not just the mean. For ML readers, the claim is useful but narrow: it rests on a few benchmark problems and a specific dependence-discovery pipeline, so independent replication on other domains would be needed before generalizing.
Sparse stochastic digital twins can outperform exact ones
arXiv:2606.16219v2 Announce Type: replace-cross Abstract: Digital twins are often built under the assumption that greater model fidelity leads to better downstream performance. We show that this is not always the case: retaining weakly dependent or auxiliary degrees of freedom can increase the dimensionality of learning and control problems enough that a sparse stochastic surrogate outperforms even an exact high-dimensional model. To identify such structure, we introduce a framework for dependence discovery directly at the level of the full conditional law. The method combines conditional generative modeling with Gaussian-process distillation and kernel mode decomposition to quantify how strongly individual variables enter the conditional distribution, including through variability, tails, multimodality, and other features invisible to the conditional mean. The resulting dependence structure can be used both to construct sparse stochastic digital twins from data and to reduce known high-dimensional models for downstream tasks. In a modified Lunar Lander problem, an eight-dimensional sparse surrogate yields better control performance than the exact 109-dimensional simulator despite lower transition fidelity. A stochastic heat-equation example shows that the framework recovers sparse structure in a 102-dimensional data-driven setting and produces a more effective surrogate than one conditioned on all candidate inputs, while economic data demonstrate its use for identifying time-varying memory structure. These results show that model fidelity and downstream utility are distinct objectives, and that dependence discovery provides a practical route to sparse stochastic digital twins.
Abstract:Digital twins are often built under the assumption that greater model fidelity leads to better downstream performance. We show that this is not always the case: retaining weakly dependent or auxiliary degrees of freedom can increase the dimensionality of learning and control problems enough that a sparse stochastic surrogate outperforms even an exact high-dimensional model. To identify such structure, we introduce a framework for dependence discovery directly at the level of the full conditional law. The method combines conditional generative modeling with Gaussian-process distillation and kernel mode decomposition to quantify how strongly individual variables enter the conditional distribution, including through variability, tails, multimodality, and other features invisible to the conditional mean. The resulting dependence structure can be used both to construct sparse stochastic digital twins from data and to reduce known high-dimensional models for downstream tasks. In a modified Lunar Lander problem, an eight-dimensional sparse surrogate yields better control performance than the exact 109-dimensional simulator despite lower transition fidelity. A stochastic heat-equation example shows that the framework recovers sparse structure in a 102-dimensional data-driven setting and produces a more effective surrogate than one conditioned on all candidate inputs, while economic data demonstrate its use for identifying time-varying memory structure. These results show that model fidelity and downstream utility are distinct objectives, and that dependence discovery provides a practical route to sparse stochastic digital twins.
Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
arXiv:2606.16219 [cs.CE]
(or arXiv:2606.16219v2 [cs.CE] for this version)
https://doi.org/10.48550/arXiv.2606.16219 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 325 cuvinte
Subiecte conexe
Expected flow networks in stochastic environments and two-player zero-sum games
arXiv · 1 sept. 2026
Diagnostic-Guided Longitudinal Modeling for Forecasting Retinal Atrophy Progression
arXiv · 22 sept. 2026
Phases of Muon: When Muon Eclipses SignSGD
arXiv · 28 oct. 2026
Revisiting Greedy Decoding for Visual Question Answering: A Calibration Perspective
arXiv · 1 sept. 2026