Metraipolis › archive › dc_fdcd1f69292a

GeoBalance: Geometry-Aware Monitoring and Reconstruction with Asymmetric Optimization for Balanced Multimodal Learning — arXiv:2609.23533

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.23533
License (as archived)
CC0-1.0
Archived
2026-10-09T21:59:41.700Z
SHA-256 of the archived text
fb7b1328e2ea59be4d6a2b7c91fb39bbafab5178bb0e51fcb28de60d5bf28080
Stable URL of this page
https://metraipolis.com/documento/dc_fdcd1f69292a

How to cite

GeoBalance: Geometry-Aware Monitoring and Reconstruction with Asymmetric Optimization for Balanced Multimodal Learning — arXiv:2609.23533 — arXiv (en), https://arxiv.org/abs/2609.23533 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fdcd1f69292a (Metraipolis archive dc_fdcd1f69292a, sha256 fb7b1328e2ea59be4d6a2b7c91fb39bbafab5178bb0e51fcb28de60d5bf28080)
BibTeX
@misc{dcfdcd1f69292a,
  title = {GeoBalance: Geometry-Aware Monitoring and Reconstruction with Asymmetric Optimization for Balanced Multimodal Learning — arXiv:2609.23533},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.23533},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fdcd1f69292a; sha256 fb7b1328e2ea59be4d6a2b7c91fb39bbafab5178bb0e51fcb28de60d5bf28080},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.23533. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal classifiers can converge to modality-dominant solutions in which one modality dominates the joint prediction, suppressing the learning of others. Existing balancing methods mainly adjust losses, gradients, or modality contributions, largely treating modality imbalance as an optimization problem while implicitly treating the weak modality as under-optimized but representationally intact. In this work, we find that this assumption does not always hold, as persistent modality dominance can induce a representation-level collapse of the weak modality, which we term \emph{manifold modality collapse} (MMC). MMC manifests as a coupled geometric degradation in which weak-modality representations collapse onto fewer directions within each class and become less separable across classes. Motivated by this observation, we propose \emph{GeoBalance}, a geometry-aware framework that monitors these two geometric properties and reconstructs the weak modality representation only when it exhibits signs of MMC. Once triggered, GeoBalance uses a fixed Simplex-ETF class scaffold and spectral regularization to restore class separation while preventing collapse onto a few feature directions. To preserve reconstruction during joint training, asymmetric gradient projection removes the joint-gradient component conflicting with reconstruction, leaving non-conflicting optimization unchanged. Extensive experiments across six multimodal benchmarks demonstrate great improvements over competitive balancing methods, validating its effectiveness.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-22 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Multimedia (cs.MM).
Pagina del lavoro: https://arxiv.org/abs/2609.23533 — 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 2026-10-09T21:45:20.914Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).