Metraipolis › archive › dc_d6ea12b0dbba
Loss-Invariant Projections as Passive Probes of Learned Representations — arXiv:2610.06195
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.06195
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:31:29.716Z
- SHA-256 of the archived text
a53cf714143ff3e9d6c77f7e0b81de2680ab6efde4bba30ea691e913af3cccfe- Stable URL of this page
https://metraipolis.com/documento/dc_d6ea12b0dbba
How to cite
Loss-Invariant Projections as Passive Probes of Learned Representations — arXiv:2610.06195 — arXiv, https://arxiv.org/abs/2610.06195 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d6ea12b0dbba (Metraipolis archive dc_d6ea12b0dbba, sha256 a53cf714143ff3e9d6c77f7e0b81de2680ab6efde4bba30ea691e913af3cccfe)
BibTeX
@misc{dcd6ea12b0dbba,
title = {Loss-Invariant Projections as Passive Probes of Learned Representations — arXiv:2610.06195},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.06195},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d6ea12b0dbba; sha256 a53cf714143ff3e9d6c77f7e0b81de2680ab6efde4bba30ea691e913af3cccfe},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.06195. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Learned feature representations in neural networks often contain structure beyond that directly used by the final task output. We study this structure using $\textit{passive probes}$ that apply fixed, untrained, property-independent projections to representations as they evolve during training. We motivate this approach through the task of prediction on $S^2$ where equivalent vector and Hermitian parameterizations reveal an additional loss-invariant trace coordinate. This motivates a general construction in which fixed random projections serve as observers of learned features. Because the observer is loss-invariant and independent of the property being studied, changes in accessibility reflect changes in the representation relative to the fixed observer rather than adaptation of the observer itself. We show that ensembles of passive probes can directly reflect task-relevant information such as target alignment. Under our constructions, the accessibility of eventual difficulty evolves differently across tasks. It increases during training in the regression tasks of surface-normal estimation and image inpainting but remains near its initial level in image classification. Comparisons with learned linear probes further show that recoverability and passive accessibility can evolve differently during training. Together, these results show how passive probes can separately characterize changes in representation geometry and the accessibility of eventual task difficulty.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.06195 — 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 2026-10-06T05:27:05.863Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).