Metraipolis › archive › dc_e758bdc1ed08
Mode Collapse Is Cheap to Detect: A Ground-Truth-Free Pre-Flight Check for Neural Samplers — arXiv:2609.26272
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.26272
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T15:07:54.178Z
- SHA-256 of the archived text
9441ba4cf912569e3c2c31a7e53993184caa6b74986554a291da4955faa75931- Stable URL of this page
https://metraipolis.com/documento/dc_e758bdc1ed08
How to cite
Mode Collapse Is Cheap to Detect: A Ground-Truth-Free Pre-Flight Check for Neural Samplers — arXiv:2609.26272 — arXiv (en), https://arxiv.org/abs/2609.26272 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e758bdc1ed08 (Metraipolis archive dc_e758bdc1ed08, sha256 9441ba4cf912569e3c2c31a7e53993184caa6b74986554a291da4955faa75931)
BibTeX
@misc{dce758bdc1ed08,
title = {Mode Collapse Is Cheap to Detect: A Ground-Truth-Free Pre-Flight Check for Neural Samplers — arXiv:2609.26272},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.26272},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e758bdc1ed08; sha256 9441ba4cf912569e3c2c31a7e53993184caa6b74986554a291da4955faa75931},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26272. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Neural samplers are trained against an unnormalised target $\tilde\pi=e^{-E}$ with no samples from $\pi$, which leaves the practitioner with no way to tell whether an expensive training run has silently dropped part of the target. The diagnostics in common use are computed from the model's own draws and are therefore confined to the model's support: we exhibit a sampler whose self-normalised effective sample size is $0.99$ while it misses $87\%$ of the target mass. We argue that \emph{detecting} missing mass is a strictly easier problem than sampling it: detection needs one point per missed basin plus a local curvature estimate, whereas correction needs the sampler retrained. We turn this into a pre-flight check that consumes a few percent of the sampler's own training budget and uses only $E$, $\nabla E$ and $\nabla^2 E$. On Gaussian-mixture, Many-Well and rotated anisotropic Many-Well targets with exactly computable ground truth, the check estimates the missing mass to within $10^{-3}$ at $2.7\%$ of training cost, where a tuned annealed SMC reference needs $70$--$280\%$ of training cost to do worse. It also applies unchanged to a controlled-SDE sampler that has no tractable density, where ESS and the ELBO cannot be formed at all. The estimator carries a \emph{self-diagnostic} that, without ground truth, is conservative in the safe direction: across $60$ configurations it clears $16$, of which $15$ are accurate to $10^{-2}$ or better. We are explicit about what this does and does not license: the check cheaply produces evidence of missing mass, and sometimes evidence that the search has stabilised, but it cannot certify a run, and its thresholds are heuristic. We then map the boundary of the method on a real physical landscape, LJ-13, and report where it fails and why.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.26272 — 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 2026-10-08T14:37:29.476Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).