Metraipolis › archive › dc_3e31cb2d1983

Predicting Unseen Scientific Interactions: Conditional Flow Models Hallucinate Non-Additive Effects — arXiv:2610.20288

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.20288
License (as archived)
CC0-1.0
Archived
2026-10-15T14:20:00.649Z
SHA-256 of the archived text
0904ca3386eef7210848478ebf1080a08b5cf4c47cf65b151fc7e06f540bc3f4
Stable URL of this page
https://metraipolis.com/documento/dc_3e31cb2d1983

How to cite

Predicting Unseen Scientific Interactions: Conditional Flow Models Hallucinate Non-Additive Effects — arXiv:2610.20288 — arXiv (en), https://arxiv.org/abs/2610.20288 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3e31cb2d1983 (Metraipolis archive dc_3e31cb2d1983, sha256 0904ca3386eef7210848478ebf1080a08b5cf4c47cf65b151fc7e06f540bc3f4)
BibTeX
@misc{dc3e31cb2d1983,
  title = {Predicting Unseen Scientific Interactions: Conditional Flow Models Hallucinate Non-Additive Effects — arXiv:2610.20288},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.20288},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3e31cb2d1983; sha256 0904ca3386eef7210848478ebf1080a08b5cf4c47cf65b151fc7e06f540bc3f4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.20288. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Conditional generative models are increasingly used for scientific prediction beyond the direct training support. Yet scientific questions often concern whether one intervention changes the effect of another. Such interactions can be of primary interest while explaining little of the total variation, allowing models to appear accurate overall without recovering the relevant scientific effect. It remains unclear how these models learn interactions and when these successfully extrapolate to unseen conditions. We therefore study interaction recovery in flow-matching models using a physiological time-series simulator, drug dose-response data, combinatorial cell perturbations, and molecular conformer generations, where we separate recovery from hallucinated interaction structure. In the controlled setting, individual effects are learned much earlier than interactions and models hallucinate interactions where none exist. Across settings, we find qualitatively different generalization problems, with successful prediction depending on whether condition structure continues from observed to held-out targets. When it does, interactions are recoverable. If not, models can generate plausible responses while missing the interaction. We furthermore find that hallucinations increase as targets become less identifiable, and that successful interaction recovery in output space can fail under nonlinear readouts. Finally, we show that aggregate fidelity metrics can conceal whether models learn and extrapolate interactions.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Quantitative Methods (q-bio.QM).
Pagina del lavoro: https://arxiv.org/abs/2610.20288 — 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 2026-10-15T13:53:35.725Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).