Online Semi-Supervised Learning on Quantized Graphs

Michal Valko, Branislav Kveton, Ling Huang, Daniel Ting
Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, PMLR R8:613-621, 2010.

Abstract

In this paper, we tackle the problem of online semi-supervised learning (SSL). When data arrive in a stream, the dual problems of com- putation and data storage arise for any SSL method. We propose a fast approximate on- line SSL algorithm that solves for the har- monic solution on an approximate graph. We show, both empirically and theoretically, that good behavior can be achieved by collapsing nearby points into a set of local “representa- tive points” that minimize distortion. More- over, we regularize the harmonic solution to achieve better stability properties. We apply our algorithm to face recognition and opti- cal character recognition applications to show that we can take advantage of the manifold structure to outperform the previous meth- ods. Unlike previous heuristic approaches, we show that our method yields provable per- formance bounds.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR8-valko10a, title = {Online Semi-Supervised Learning on Quantized Graphs}, author = {Valko, Michal and Kveton, Branislav and Huang, Ling and Ting, Daniel}, booktitle = {Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence}, pages = {613--621}, 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/valko10a/valko10a.pdf}, url = {https://proceedings.mlr.press/r8/valko10a.html}, abstract = {In this paper, we tackle the problem of online semi-supervised learning (SSL). When data arrive in a stream, the dual problems of com- putation and data storage arise for any SSL method. We propose a fast approximate on- line SSL algorithm that solves for the har- monic solution on an approximate graph. We show, both empirically and theoretically, that good behavior can be achieved by collapsing nearby points into a set of local “representa- tive points” that minimize distortion. More- over, we regularize the harmonic solution to achieve better stability properties. We apply our algorithm to face recognition and opti- cal character recognition applications to show that we can take advantage of the manifold structure to outperform the previous meth- ods. Unlike previous heuristic approaches, we show that our method yields provable per- formance bounds.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Online Semi-Supervised Learning on Quantized Graphs %A Michal Valko %A Branislav Kveton %A Ling Huang %A Daniel Ting %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-valko10a %I PMLR %P 613--621 %U https://proceedings.mlr.press/r8/valko10a.html %V R8 %X In this paper, we tackle the problem of online semi-supervised learning (SSL). When data arrive in a stream, the dual problems of com- putation and data storage arise for any SSL method. We propose a fast approximate on- line SSL algorithm that solves for the har- monic solution on an approximate graph. We show, both empirically and theoretically, that good behavior can be achieved by collapsing nearby points into a set of local “representa- tive points” that minimize distortion. More- over, we regularize the harmonic solution to achieve better stability properties. We apply our algorithm to face recognition and opti- cal character recognition applications to show that we can take advantage of the manifold structure to outperform the previous meth- ods. Unlike previous heuristic approaches, we show that our method yields provable per- formance bounds. %Z Reissued by PMLR on 04 October 2026.
APA
Valko, M., Kveton, B., Huang, L. & Ting, D.. (2010). Online Semi-Supervised Learning on Quantized Graphs. Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R8:613-621 Available from https://proceedings.mlr.press/r8/valko10a.html. Reissued by PMLR on 04 October 2026.

Related Material