Metraipolis › archive › dc_be78f8efc175

Spectral Feedback for Test-Time Alignment of Protein Diffusion Models — arXiv:2609.30456

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.30456
License (as archived)
CC0-1.0
Archived
2026-10-08T01:38:08.955Z
SHA-256 of the archived text
341bc5385f18fbdc2d5dc480d181533ea26aab03eb407d68d8058fad5ad65f61
Stable URL of this page
https://metraipolis.com/documento/dc_be78f8efc175

How to cite

Spectral Feedback for Test-Time Alignment of Protein Diffusion Models — arXiv:2609.30456 — arXiv (en), https://arxiv.org/abs/2609.30456 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_be78f8efc175 (Metraipolis archive dc_be78f8efc175, sha256 341bc5385f18fbdc2d5dc480d181533ea26aab03eb407d68d8058fad5ad65f61)
BibTeX
@misc{dcbe78f8efc175,
  title = {Spectral Feedback for Test-Time Alignment of Protein Diffusion Models — arXiv:2609.30456},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30456},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_be78f8efc175; sha256 341bc5385f18fbdc2d5dc480d181533ea26aab03eb407d68d8058fad5ad65f61},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30456. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reward maximization alignment methods for discrete diffusion models have primarily focused on steering the reverse process, either by influencing token logits or by selecting favorable sequences at intermediate steps. These approaches largely treat inference as a unidirectional process, lacking mechanisms for revisiting undesirable token selections. We introduce Spectral Feedback, an algorithm that selects edit-positions in a feedback loop, allowing the model to iteratively correct its own generations. This approach leverages the mask structure of discrete diffusion models by re-masking and re-sampling tokens, analogous to image editing methods that reintroduce noisy latents and re-run the reverse process. While prior alignment methods focus on what token labels to assign to maximize a target reward, we instead treat which tokens to revisit as the central alignment problem. Selecting edit-positions is challenging because edit effects are interdependent: the impact of modifying one token depends on which others are edited simultaneously. We define an edit-set as a set of token positions to re-mask and re-sample. Motivated by prior work on sparse interactions in biological systems, we find empirically that edit-set value functions for protein inverse folding admit sparse Fourier representations. This structure enables Spectral Feedback to efficiently learn and optimize the value functions for edit-position selection. Spectral Feedback is model-agnostic and can be applied to pretrained, test-time aligned, and fine-tuned diffusion models. For all of these models, the algorithm improves alignment performance without modifying the underlying generative process. Applied to inverse folding with a protein stability reward oracle, it achieves a 32.3% increase in stable proteins for a pretrained model, 24.8% for Best-of-10, and 5.8% for a state-of-the-art RL fine-tuned diffusion model.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.30456 — 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 2026-10-08T01:12:14.238Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).