An Online Learning-based Framework for Tracking

Kamalika Chaudhuri, Yoav Freund, Daniel Hsu
Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, PMLR R8:109-116, 2010.

Abstract

We study the tracking problem, namely, es- timating the hidden state of an object over time, from unreliable and noisy measure- ments. The standard framework for the tracking problem is the generative frame- work, which is the basis of solutions such as the Bayesian algorithm and its approxima- tion, the particle filters. However, these so- lutions can be very sensitive to model mis- matches. In this paper, motivated by online learning, we introduce a new framework for tracking. We provide an efficient tracking al- gorithm for this framework. We provide ex- perimental results comparing our algorithm to the Bayesian algorithm on simulated data. Our experiments show that when there are slight model mismatches, our algorithm out- performs the Bayesian algorithm.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR8-chaudhuri10a, title = {An Online Learning-based Framework for Tracking}, author = {Chaudhuri, Kamalika and Freund, Yoav and Hsu, Daniel}, booktitle = {Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence}, pages = {109--116}, 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/chaudhuri10a/chaudhuri10a.pdf}, url = {https://proceedings.mlr.press/r8/chaudhuri10a.html}, abstract = {We study the tracking problem, namely, es- timating the hidden state of an object over time, from unreliable and noisy measure- ments. The standard framework for the tracking problem is the generative frame- work, which is the basis of solutions such as the Bayesian algorithm and its approxima- tion, the particle filters. However, these so- lutions can be very sensitive to model mis- matches. In this paper, motivated by online learning, we introduce a new framework for tracking. We provide an efficient tracking al- gorithm for this framework. We provide ex- perimental results comparing our algorithm to the Bayesian algorithm on simulated data. Our experiments show that when there are slight model mismatches, our algorithm out- performs the Bayesian algorithm.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T An Online Learning-based Framework for Tracking %A Kamalika Chaudhuri %A Yoav Freund %A Daniel Hsu %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-chaudhuri10a %I PMLR %P 109--116 %U https://proceedings.mlr.press/r8/chaudhuri10a.html %V R8 %X We study the tracking problem, namely, es- timating the hidden state of an object over time, from unreliable and noisy measure- ments. The standard framework for the tracking problem is the generative frame- work, which is the basis of solutions such as the Bayesian algorithm and its approxima- tion, the particle filters. However, these so- lutions can be very sensitive to model mis- matches. In this paper, motivated by online learning, we introduce a new framework for tracking. We provide an efficient tracking al- gorithm for this framework. We provide ex- perimental results comparing our algorithm to the Bayesian algorithm on simulated data. Our experiments show that when there are slight model mismatches, our algorithm out- performs the Bayesian algorithm. %Z Reissued by PMLR on 04 October 2026.
APA
Chaudhuri, K., Freund, Y. & Hsu, D.. (2010). An Online Learning-based Framework for Tracking. Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R8:109-116 Available from https://proceedings.mlr.press/r8/chaudhuri10a.html. Reissued by PMLR on 04 October 2026.

Related Material