Probabilistic AND-OR Attribute Grouping for Zero-Shot Learning

Yuval Atzmon, Gal Chechik
Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, PMLR R16:381-391, 2018.

Abstract

In zero-shot learning (ZSL), a classifier is trained to recognize visual classes without any image samples. Instead, it is given semantic information about the class, like a textual description or a set of attributes. Learning from attributes could benefit from explicitly modeling structure of the attribute space. Unfortunately, learning of general structure from empirical samples is hard with typical dataset sizes. Here we describe LAGO1, a probabilistic model designed to capture natural soft and or relations across groups of attributes. We show how this model can be learned end-to-end with a deep attribute-detection model. The soft group structure can be learned from data jointly as part of the model, and can also readily incorporate prior knowledge about groups if available. The soft and-or structure succeeds to capture meaningful and predictive structures, improving the accuracy of zero-shot learning on two of three benchmarks. Finally, LAGO reveals a unified formulation over two ZSL approaches: DAP (Lampert et al., 2009) and ESZSL (Romera-Paredes & Torr, 2015). Interestingly, taking only one singleton group for each attribute, introduces a new soft-relaxation of DAP, that outperforms DAP by $\tilde$40%.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR16-atzmon18a, title = {Probabilistic {AND}-{OR} Attribute Grouping for Zero-Shot Learning}, author = {Atzmon, Yuval and Chechik, Gal}, booktitle = {Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence}, pages = {381--391}, year = {2018}, editor = {Globerson, Amir and Silva, Ricardo}, volume = {R16}, series = {Proceedings of Machine Learning Research}, month = {06--10 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r16/main/assets/atzmon18a/atzmon18a.pdf}, url = {https://proceedings.mlr.press/r16/atzmon18a.html}, abstract = {In zero-shot learning (ZSL), a classifier is trained to recognize visual classes without any image samples. Instead, it is given semantic information about the class, like a textual description or a set of attributes. Learning from attributes could benefit from explicitly modeling structure of the attribute space. Unfortunately, learning of general structure from empirical samples is hard with typical dataset sizes. Here we describe LAGO1, a probabilistic model designed to capture natural soft and or relations across groups of attributes. We show how this model can be learned end-to-end with a deep attribute-detection model. The soft group structure can be learned from data jointly as part of the model, and can also readily incorporate prior knowledge about groups if available. The soft and-or structure succeeds to capture meaningful and predictive structures, improving the accuracy of zero-shot learning on two of three benchmarks. Finally, LAGO reveals a unified formulation over two ZSL approaches: DAP (Lampert et al., 2009) and ESZSL (Romera-Paredes & Torr, 2015). Interestingly, taking only one singleton group for each attribute, introduces a new soft-relaxation of DAP, that outperforms DAP by $\tilde$40%.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Probabilistic AND-OR Attribute Grouping for Zero-Shot Learning %A Yuval Atzmon %A Gal Chechik %B Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2018 %E Amir Globerson %E Ricardo Silva %F pmlr-vR16-atzmon18a %I PMLR %P 381--391 %U https://proceedings.mlr.press/r16/atzmon18a.html %V R16 %X In zero-shot learning (ZSL), a classifier is trained to recognize visual classes without any image samples. Instead, it is given semantic information about the class, like a textual description or a set of attributes. Learning from attributes could benefit from explicitly modeling structure of the attribute space. Unfortunately, learning of general structure from empirical samples is hard with typical dataset sizes. Here we describe LAGO1, a probabilistic model designed to capture natural soft and or relations across groups of attributes. We show how this model can be learned end-to-end with a deep attribute-detection model. The soft group structure can be learned from data jointly as part of the model, and can also readily incorporate prior knowledge about groups if available. The soft and-or structure succeeds to capture meaningful and predictive structures, improving the accuracy of zero-shot learning on two of three benchmarks. Finally, LAGO reveals a unified formulation over two ZSL approaches: DAP (Lampert et al., 2009) and ESZSL (Romera-Paredes & Torr, 2015). Interestingly, taking only one singleton group for each attribute, introduces a new soft-relaxation of DAP, that outperforms DAP by $\tilde$40%. %Z Reissued by PMLR on 04 October 2026.
APA
Atzmon, Y. & Chechik, G.. (2018). Probabilistic AND-OR Attribute Grouping for Zero-Shot Learning. Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R16:381-391 Available from https://proceedings.mlr.press/r16/atzmon18a.html. Reissued by PMLR on 04 October 2026.

Related Material