Spectral Chinese Restaurant Processes: Nonparametric Clustering Based on Similarities

Richard Socher, Andrew Maas, Christopher Manning
Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, PMLR 15:698-706, 2011.

Abstract

We introduce a new nonparametric clustering model which combines the recently proposed distance-dependent Chinese restaurant process (dd-CRP) and non-linear, spectral methods for dimensionality reduction. Our model retains the ability of nonparametric methods to learn the number of clusters from data. At the same time it addresses two key limitations of nonparametric Bayesian methods: modeling data that are not exchangeable and have many correlated features. Spectral methods use the similarity between documents to map them into a low-dimensional spectral space where we then compare several clustering methods. Our experiments on handwritten digits and text documents show that nonparametric methods such as the CRP or dd-CRP can perform as well as or better than $k$-means and also recover the true number of clusters. We improve the performance of the dd-CRP in spectral space by incorporating the original similarity matrix in its prior. This simple modification results in better performance than all other methods we compared to. We offer a new formulation and first experimental evaluation of a general Gibbs sampler for mixture modeling with distance-dependent CRPs.

Cite this Paper


BibTeX
@InProceedings{pmlr-v15-socher11a, title = {Spectral Chinese Restaurant Processes: Nonparametric Clustering Based on Similarities}, author = {Socher, Richard and Maas, Andrew and Manning, Christopher}, booktitle = {Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics}, pages = {698--706}, year = {2011}, editor = {Gordon, Geoffrey and Dunson, David and Dudík, Miroslav}, volume = {15}, series = {Proceedings of Machine Learning Research}, address = {Fort Lauderdale, FL, USA}, month = {11--13 Apr}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v15/socher11a/socher11a.pdf}, url = {https://proceedings.mlr.press/v15/socher11a.html}, abstract = {We introduce a new nonparametric clustering model which combines the recently proposed distance-dependent Chinese restaurant process (dd-CRP) and non-linear, spectral methods for dimensionality reduction. Our model retains the ability of nonparametric methods to learn the number of clusters from data. At the same time it addresses two key limitations of nonparametric Bayesian methods: modeling data that are not exchangeable and have many correlated features. Spectral methods use the similarity between documents to map them into a low-dimensional spectral space where we then compare several clustering methods. Our experiments on handwritten digits and text documents show that nonparametric methods such as the CRP or dd-CRP can perform as well as or better than $k$-means and also recover the true number of clusters. We improve the performance of the dd-CRP in spectral space by incorporating the original similarity matrix in its prior. This simple modification results in better performance than all other methods we compared to. We offer a new formulation and first experimental evaluation of a general Gibbs sampler for mixture modeling with distance-dependent CRPs.} }
Endnote
%0 Conference Paper %T Spectral Chinese Restaurant Processes: Nonparametric Clustering Based on Similarities %A Richard Socher %A Andrew Maas %A Christopher Manning %B Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2011 %E Geoffrey Gordon %E David Dunson %E Miroslav Dudík %F pmlr-v15-socher11a %I PMLR %P 698--706 %U https://proceedings.mlr.press/v15/socher11a.html %V 15 %X We introduce a new nonparametric clustering model which combines the recently proposed distance-dependent Chinese restaurant process (dd-CRP) and non-linear, spectral methods for dimensionality reduction. Our model retains the ability of nonparametric methods to learn the number of clusters from data. At the same time it addresses two key limitations of nonparametric Bayesian methods: modeling data that are not exchangeable and have many correlated features. Spectral methods use the similarity between documents to map them into a low-dimensional spectral space where we then compare several clustering methods. Our experiments on handwritten digits and text documents show that nonparametric methods such as the CRP or dd-CRP can perform as well as or better than $k$-means and also recover the true number of clusters. We improve the performance of the dd-CRP in spectral space by incorporating the original similarity matrix in its prior. This simple modification results in better performance than all other methods we compared to. We offer a new formulation and first experimental evaluation of a general Gibbs sampler for mixture modeling with distance-dependent CRPs.
RIS
TY - CPAPER TI - Spectral Chinese Restaurant Processes: Nonparametric Clustering Based on Similarities AU - Richard Socher AU - Andrew Maas AU - Christopher Manning BT - Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics DA - 2011/06/14 ED - Geoffrey Gordon ED - David Dunson ED - Miroslav Dudík ID - pmlr-v15-socher11a PB - PMLR DP - Proceedings of Machine Learning Research VL - 15 SP - 698 EP - 706 L1 - http://proceedings.mlr.press/v15/socher11a/socher11a.pdf UR - https://proceedings.mlr.press/v15/socher11a.html AB - We introduce a new nonparametric clustering model which combines the recently proposed distance-dependent Chinese restaurant process (dd-CRP) and non-linear, spectral methods for dimensionality reduction. Our model retains the ability of nonparametric methods to learn the number of clusters from data. At the same time it addresses two key limitations of nonparametric Bayesian methods: modeling data that are not exchangeable and have many correlated features. Spectral methods use the similarity between documents to map them into a low-dimensional spectral space where we then compare several clustering methods. Our experiments on handwritten digits and text documents show that nonparametric methods such as the CRP or dd-CRP can perform as well as or better than $k$-means and also recover the true number of clusters. We improve the performance of the dd-CRP in spectral space by incorporating the original similarity matrix in its prior. This simple modification results in better performance than all other methods we compared to. We offer a new formulation and first experimental evaluation of a general Gibbs sampler for mixture modeling with distance-dependent CRPs. ER -
APA
Socher, R., Maas, A. & Manning, C.. (2011). Spectral Chinese Restaurant Processes: Nonparametric Clustering Based on Similarities. Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 15:698-706 Available from https://proceedings.mlr.press/v15/socher11a.html.

Related Material