Hybrid Generative/Discriminative Learning for Automatic Image Annotation

Shuang-Hong Yang, Jiang Bian, Hongyuan Zha
Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, PMLR R8:682-689, 2010.

Abstract

Automatic image annotation (AIA) raises tremendous challenges to machine learning as it requires modeling of data that are both ambiguous in input and output, e.g., images containing multiple objects and labeled with multiple semantic tags. Even more challenging is that the number of candidate tags is usually huge (as large as the vocabulary size) yet each image is only related to a few of them. This pa- per presents a hybrid generative-discriminative classifier to simultaneously address the extreme data-ambiguity and overfitting-vulnerability issues in tasks such as AIA. Particularly: (1) an Exponential-Multinomial Mixture (EMM) model is established to capture both the input and output ambiguity and in the meanwhile to encourage prediction sparsity; and (2) the prediction ability of the EMM model is explicitly maximized through discriminative learning that integrates variational inference of graphical models and the pairwise formulation of ordinal regression. Experiments show that our approach achieves both superior annotation performance and better tag scalability.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR8-yang10a, title = {Hybrid Generative/Discriminative Learning for Automatic Image Annotation}, author = {Yang, Shuang-Hong and Bian, Jiang and Zha, Hongyuan}, booktitle = {Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence}, pages = {682--689}, year = {2010}, editor = {Grünwald, Peter and Spirtes, Peter}, volume = {R8}, series = {Proceedings of Machine Learning Research}, month = {08--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r8/main/assets/yang10a/yang10a.pdf}, url = {https://proceedings.mlr.press/r8/yang10a.html}, abstract = {Automatic image annotation (AIA) raises tremendous challenges to machine learning as it requires modeling of data that are both ambiguous in input and output, e.g., images containing multiple objects and labeled with multiple semantic tags. Even more challenging is that the number of candidate tags is usually huge (as large as the vocabulary size) yet each image is only related to a few of them. This pa- per presents a hybrid generative-discriminative classifier to simultaneously address the extreme data-ambiguity and overfitting-vulnerability issues in tasks such as AIA. Particularly: (1) an Exponential-Multinomial Mixture (EMM) model is established to capture both the input and output ambiguity and in the meanwhile to encourage prediction sparsity; and (2) the prediction ability of the EMM model is explicitly maximized through discriminative learning that integrates variational inference of graphical models and the pairwise formulation of ordinal regression. Experiments show that our approach achieves both superior annotation performance and better tag scalability.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Hybrid Generative/Discriminative Learning for Automatic Image Annotation %A Shuang-Hong Yang %A Jiang Bian %A Hongyuan Zha %B Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2010 %E Peter Grünwald %E Peter Spirtes %F pmlr-vR8-yang10a %I PMLR %P 682--689 %U https://proceedings.mlr.press/r8/yang10a.html %V R8 %X Automatic image annotation (AIA) raises tremendous challenges to machine learning as it requires modeling of data that are both ambiguous in input and output, e.g., images containing multiple objects and labeled with multiple semantic tags. Even more challenging is that the number of candidate tags is usually huge (as large as the vocabulary size) yet each image is only related to a few of them. This pa- per presents a hybrid generative-discriminative classifier to simultaneously address the extreme data-ambiguity and overfitting-vulnerability issues in tasks such as AIA. Particularly: (1) an Exponential-Multinomial Mixture (EMM) model is established to capture both the input and output ambiguity and in the meanwhile to encourage prediction sparsity; and (2) the prediction ability of the EMM model is explicitly maximized through discriminative learning that integrates variational inference of graphical models and the pairwise formulation of ordinal regression. Experiments show that our approach achieves both superior annotation performance and better tag scalability. %Z Reissued by PMLR on 04 October 2026.
APA
Yang, S., Bian, J. & Zha, H.. (2010). Hybrid Generative/Discriminative Learning for Automatic Image Annotation. Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R8:682-689 Available from https://proceedings.mlr.press/r8/yang10a.html. Reissued by PMLR on 04 October 2026.

Related Material