Metraipolis › archive › dc_e491ef0c8577

GradRepair-ODE: Certified Gradient Repair for Neural ODE Training — arXiv:2609.13204

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.13204
License (as archived)
CC0-1.0
Archived
2026-10-15T23:16:37.573Z
SHA-256 of the archived text
d3e08339e06840711f42267f7f5913105b15dc0c0dbc88edc4956ab58fa05d73
Stable URL of this page
https://metraipolis.com/documento/dc_e491ef0c8577

How to cite

GradRepair-ODE: Certified Gradient Repair for Neural ODE Training — arXiv:2609.13204 — arXiv (en), https://arxiv.org/abs/2609.13204 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e491ef0c8577 (Metraipolis archive dc_e491ef0c8577, sha256 d3e08339e06840711f42267f7f5913105b15dc0c0dbc88edc4956ab58fa05d73)
BibTeX
@misc{dce491ef0c8577,
  title = {GradRepair-ODE: Certified Gradient Repair for Neural ODE Training — arXiv:2609.13204},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.13204},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e491ef0c8577; sha256 d3e08339e06840711f42267f7f5913105b15dc0c0dbc88edc4956ab58fa05d73},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.13204. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Neural ordinary differential equations use numerical solvers inside the training loop. The solver determines the forward trajectory and also affects the gradient passed to the optimizer. That coupling creates a reliability problem for scientific machine learning and continuous-time generative modeling, including diffusion probability-flow ordinary differential equations and flow-matching models. Under loose step sizes, stiff dynamics, chaotic sensitivity, or event discontinuities, a differentiable ODE pipeline can return a finite gradient whose direction is numerically suspect. We introduce GradRepair-ODE, a reliability framework for checking, repairing, and rejecting ODE gradients at the optimizer step. The method computes several gradient candidates, compares them with directional finite-difference checks and solver diagnostics, diagnoses likely numerical failure modes, repairs selected gradients through path switching or stricter recomputation, and rejects steps whose descent direction cannot be certified. In six synthetic ODE systems, GradRepair-ODE leaves low-risk systems unchanged, repairs Robertson and Lorenz gradients to cosine similarity 1.000 against a strict reference, reduces unsafe accepted steps from 37 to 0, and rejects an event-discontinuous case instead of applying an uncertified update. The paper argues for a simple change in the training contract: an ODE gradient should reach the optimizer with numerical evidence attached.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.CL).
Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.13204 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T22:51:09.988Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).