aiqbench analysis
This paper presents a rigorous finite-sample theory for geometrically supervised latent world models, establishing conditions under which learned controllers maintain metric faithfulness without collapsing into degenerate representations. By proving that approximate empirical minimizers are simultaneously co-Lipschitz and semiconjugate to controlled dynamics, the work provides explicit error bounds and deterministic guarantees essential for reliable planning in nonlinear systems.
Finite-Sample Metric Non-Collapse for Geometrically Supervised Latent World Models in Control
arXiv:2608.07265v2 Announce Type: cross Abstract: We establish a finite-sample learning-to-control theory for geometrically supervised latent models of nonlinear deterministic systems. Geometric supervision is used only during training: simulator state, proprioception, or state estimates with independently validated metric and directional error bounds supply observable-state distances and tangent directions, while deployment remains observation- and action-conditioned. We introduce an encoder-only local--global metric hinge that enforces directional resolution and separated-state discrimination. Under regular observable-factor, coverage, finite-capacity approximation, and uniform $C^{1,1}$ hypotheses, a computable one-sided regularization regime has a strong selection property: with high probability, every approximate empirical minimizer is simultaneously pointwise co-Lipschitz and uniformly approximately semiconjugate to the controlled dynamics. Approximation, sampling, and optimization errors remain explicit and separate. Norm-constrained tensor-product B-spline classes constructively realize the approximation hypotheses, and the interpolation exponent converting mean residual control into a uniform bound is sharp. A modular deterministic corollary transfers the learned certificates to trajectory, finite-horizon cost, learned-cost-head, and optimizer guarantees, while a validated finite-net result enables sharper model-specific certification. Controlled experiments isolate collapse and folding, quantify the analytic certificate's reserve, and demonstrate the control benefit of restored metric resolution. The principal contribution is a complete finite-sample implication from approximate empirical optimization to metric faithfulness, uniform controlled dynamics, and reliable planning for the same learned model.
Abstract:We establish a finite-sample learning-to-control theory for geometrically supervised latent models of nonlinear deterministic systems. Geometric supervision is used only during training: simulator state, proprioception, or state estimates with independently validated metric and directional error bounds supply observable-state distances and tangent directions, while deployment remains observation- and action-conditioned. We introduce an encoder-only local--global metric hinge that enforces directional resolution and separated-state discrimination. Under regular observable-factor, coverage, finite-capacity approximation, and uniform $C^{1,1}$ hypotheses, a computable one-sided regularization regime has a strong selection property: with high probability, every approximate empirical minimizer is simultaneously pointwise co-Lipschitz and uniformly approximately semiconjugate to the controlled dynamics. Approximation, sampling, and optimization errors remain explicit and separate. Norm-constrained tensor-product B-spline classes constructively realize the approximation hypotheses, and the interpolation exponent converting mean residual control into a uniform bound is sharp. A modular deterministic corollary transfers the learned certificates to trajectory, finite-horizon cost, learned-cost-head, and optimizer guarantees, while a validated finite-net result enables sharper model-specific certification. Controlled experiments isolate collapse and folding, quantify the analytic certificate's reserve, and demonstrate the control benefit of restored metric resolution. The principal contribution is a complete finite-sample implication from approximate empirical optimization to metric faithfulness, uniform controlled dynamics, and reliable planning for the same learned model.
Revised version prepared in response to the editorial assessment. The main manuscript is 32 pages; detailed mathematical derivations have been moved to the accompanying Supplementary Material. The principal results and contributions are strengthened and clarified
Optimization and Control (math.OC); Machine Learning (cs.LG)
arXiv:2608.07265 [math.OC]
(or arXiv:2608.07265v2 [math.OC] for this version)
Full text under the publisher's open licence · 254 words
Related coverage
GDLAM: Group-Disentangled Latent Action Model for Highly Disentangled Embodied Pretraining
arXiv · Sep 24, 2026
Exploring self-distilled reasoning for supervised fine-tuning with Amazon Nova
AWS · Jul 21, 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · Oct 13, 2026
[AINews] not much happened today
Latent Space · Sep 10, 2026