Abstract | ||
---|---|---|
Principal component analysis (PCA) has been extensively applied in data mining, pattern recognition and information retrieval for unsupervised dimensionality reduction. When labels of data are available, e.g., in a classification or regres- sion task, PCA is however not able to use this information. The problem is more interesting if only part of the input data are labeled, i.e., in a semi-supervised setting. In this paper we propose a supervised PCA model called SPPCA and a semi-supervised PCA model called S2PPCA, both of which are extensions of a probabilistic PCA model. The pro- posed models are able to incorporate the label information into the projection phase, and can naturally handle multi- ple outputs (i.e., in multi-task learning problems). We de- rive an ecient EM learning algorithm for both models, and also provide theoretical justifications of the model behaviors. SPPCA and S2PPCA are compared with other supervised projection methods on various learning tasks, and show not only promising performance but also good scalability. |
Year | DOI | Venue |
---|---|---|
2006 | 10.1145/1150402.1150454 | KDD |
Keywords | Field | DocType |
principal component analysis,data mining,probabilistic pca model,input data,various learning task,su- pervised projection,model behavior,information retrieval,supervised pca model,supervised probabilistic principal component,semi-supervised pca model,semi-supervised projection,label information,dimensionality reduction,projection method,pattern recognition,multi task learning | Data mining,Dimensionality reduction,Semi-supervised learning,Computer science,Artificial intelligence,Probabilistic principal component analysis,Probabilistic logic,Sparse PCA,Regression,Pattern recognition,Machine learning,Principal component analysis,Scalability | Conference |
ISBN | Citations | PageRank |
1-59593-339-5 | 35 | 2.56 |
References | Authors | |
8 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Shipeng Yu | 1 | 1767 | 118.84 |
Yu, Kai | 2 | 4799 | 255.21 |
Volker Tresp | 3 | 2907 | 373.75 |
Hans-Peter Kriegel | 4 | 20742 | 3284.07 |
Mingrui Wu | 5 | 515 | 23.03 |