AClass: A simple, online, parallelizable algorithm for probabilistic classification

Vikash K. Mansinghka, Daniel M. Roy, Ryan Rifkin, Josh Tenenbaum
Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, PMLR 2:315-322, 2007.

Abstract

We present AClass, a simple, online, parallelizable algorithm for supervised multiclass classification. AClass models each classconditional density as a Chinese restaurant process mixture, and performs approximate inference in this model using a sequential Monte Carlo scheme. AClass combines several strengths of previous approaches to classification that are not typically found in a single algorithm; it supports learning from missing data and yields sensibly regularized nonlinear decision boundaries while remaining computationally efficient. We compare AClass to several standard classification algorithms and show competitive performance.

Cite this Paper


BibTeX
@InProceedings{pmlr-v2-mansinghka07a, title = {AClass: A simple, online, parallelizable algorithm for probabilistic classification}, author = {Mansinghka, Vikash K. and Roy, Daniel M. and Rifkin, Ryan and Tenenbaum, Josh}, booktitle = {Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics}, pages = {315--322}, year = {2007}, editor = {Meila, Marina and Shen, Xiaotong}, volume = {2}, series = {Proceedings of Machine Learning Research}, address = {San Juan, Puerto Rico}, month = {21--24 Mar}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v2/mansinghka07a/mansinghka07a.pdf}, url = {https://proceedings.mlr.press/v2/mansinghka07a.html}, abstract = {We present AClass, a simple, online, parallelizable algorithm for supervised multiclass classification. AClass models each classconditional density as a Chinese restaurant process mixture, and performs approximate inference in this model using a sequential Monte Carlo scheme. AClass combines several strengths of previous approaches to classification that are not typically found in a single algorithm; it supports learning from missing data and yields sensibly regularized nonlinear decision boundaries while remaining computationally efficient. We compare AClass to several standard classification algorithms and show competitive performance.} }
Endnote
%0 Conference Paper %T AClass: A simple, online, parallelizable algorithm for probabilistic classification %A Vikash K. Mansinghka %A Daniel M. Roy %A Ryan Rifkin %A Josh Tenenbaum %B Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2007 %E Marina Meila %E Xiaotong Shen %F pmlr-v2-mansinghka07a %I PMLR %P 315--322 %U https://proceedings.mlr.press/v2/mansinghka07a.html %V 2 %X We present AClass, a simple, online, parallelizable algorithm for supervised multiclass classification. AClass models each classconditional density as a Chinese restaurant process mixture, and performs approximate inference in this model using a sequential Monte Carlo scheme. AClass combines several strengths of previous approaches to classification that are not typically found in a single algorithm; it supports learning from missing data and yields sensibly regularized nonlinear decision boundaries while remaining computationally efficient. We compare AClass to several standard classification algorithms and show competitive performance.
RIS
TY - CPAPER TI - AClass: A simple, online, parallelizable algorithm for probabilistic classification AU - Vikash K. Mansinghka AU - Daniel M. Roy AU - Ryan Rifkin AU - Josh Tenenbaum BT - Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics DA - 2007/03/11 ED - Marina Meila ED - Xiaotong Shen ID - pmlr-v2-mansinghka07a PB - PMLR DP - Proceedings of Machine Learning Research VL - 2 SP - 315 EP - 322 L1 - http://proceedings.mlr.press/v2/mansinghka07a/mansinghka07a.pdf UR - https://proceedings.mlr.press/v2/mansinghka07a.html AB - We present AClass, a simple, online, parallelizable algorithm for supervised multiclass classification. AClass models each classconditional density as a Chinese restaurant process mixture, and performs approximate inference in this model using a sequential Monte Carlo scheme. AClass combines several strengths of previous approaches to classification that are not typically found in a single algorithm; it supports learning from missing data and yields sensibly regularized nonlinear decision boundaries while remaining computationally efficient. We compare AClass to several standard classification algorithms and show competitive performance. ER -
APA
Mansinghka, V.K., Roy, D.M., Rifkin, R. & Tenenbaum, J.. (2007). AClass: A simple, online, parallelizable algorithm for probabilistic classification. Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 2:315-322 Available from https://proceedings.mlr.press/v2/mansinghka07a.html.

Related Material