Learning Visual-Semantic Subspace Representations

Gabriel Moreira, Manuel Marques, Joao Costeira, Alexander G Hauptmann
Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, PMLR 258:3727-3735, 2025.

Abstract

Learning image representations that capture rich semantic relationships remains a significant challenge. Existing approaches are either contrastive, lacking robust theoretical guarantees, or struggle to effectively represent the partial orders inherent to structured visual-semantic data. In this paper, we introduce a nuclear norm-based loss function, grounded in the same information theoretic principles that have proved effective in self-supervised learning. We present a theoretical characterization of this loss, demonstrating that, in addition to promoting class orthogonality, it encodes the spectral geometry of the data within a subspace lattice. This geometric representation allows us to associate logical propositions with subspaces, ensuring that our learned representations adhere to a predefined symbolic structure.

Cite this Paper

BibTeX
@InProceedings{pmlr-v258-moreira25a, title = {Learning Visual-Semantic Subspace Representations}, author = {Moreira, Gabriel and Marques, Manuel and Costeira, Joao and Hauptmann, Alexander G}, booktitle = {Proceedings of The 28th International Conference on Artificial Intelligence and Statistics}, pages = {3727--3735}, year = {2025}, editor = {Li, Yingzhen and Mandt, Stephan and Agrawal, Shipra and Khan, Emtiyaz}, volume = {258}, series = {Proceedings of Machine Learning Research}, month = {03--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v258/main/assets/moreira25a/moreira25a.pdf}, url = {https://proceedings.mlr.press/v258/moreira25a.html}, abstract = {Learning image representations that capture rich semantic relationships remains a significant challenge. Existing approaches are either contrastive, lacking robust theoretical guarantees, or struggle to effectively represent the partial orders inherent to structured visual-semantic data. In this paper, we introduce a nuclear norm-based loss function, grounded in the same information theoretic principles that have proved effective in self-supervised learning. We present a theoretical characterization of this loss, demonstrating that, in addition to promoting class orthogonality, it encodes the spectral geometry of the data within a subspace lattice. This geometric representation allows us to associate logical propositions with subspaces, ensuring that our learned representations adhere to a predefined symbolic structure.} }
Endnote
%0 Conference Paper %T Learning Visual-Semantic Subspace Representations %A Gabriel Moreira %A Manuel Marques %A Joao Costeira %A Alexander G Hauptmann %B Proceedings of The 28th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2025 %E Yingzhen Li %E Stephan Mandt %E Shipra Agrawal %E Emtiyaz Khan %F pmlr-v258-moreira25a %I PMLR %P 3727--3735 %U https://proceedings.mlr.press/v258/moreira25a.html %V 258 %X Learning image representations that capture rich semantic relationships remains a significant challenge. Existing approaches are either contrastive, lacking robust theoretical guarantees, or struggle to effectively represent the partial orders inherent to structured visual-semantic data. In this paper, we introduce a nuclear norm-based loss function, grounded in the same information theoretic principles that have proved effective in self-supervised learning. We present a theoretical characterization of this loss, demonstrating that, in addition to promoting class orthogonality, it encodes the spectral geometry of the data within a subspace lattice. This geometric representation allows us to associate logical propositions with subspaces, ensuring that our learned representations adhere to a predefined symbolic structure.
APA
Moreira, G., Marques, M., Costeira, J. & Hauptmann, A.G.. (2025). Learning Visual-Semantic Subspace Representations. Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 258:3727-3735 Available from https://proceedings.mlr.press/v258/moreira25a.html.

Related Material