论文信息 - Semi-supervised Learning via Gaussian Processes

Semi-supervised Learning via Gaussian Processes

We present a probabilistic approach to learning a Gaussian Process classifier in the presence of unlabeled data. Our approach involves a "null category noise model" (NCNM) inspired by ordered categorical noise models. The noise model reflects an assumption that the data density is lower between the class-conditional densities. We illustrate our approach on a toy problem and present comparative results for the semi-supervised classification of handwritten digits.

Neil D. Lawrence | Michael I. Jordan | Neil D. Lawrence

[1] Alan Agresti,et al. Categorical Data Analysis , 1991, International Encyclopedia of Statistical Science.

[2] Christopher K. I. Williams. Prediction with Gaussian Processes: From Linear Regression to Linear Prediction and Beyond , 1999, Learning in Graphical Models.

[3] Vladimir Vapnik,et al. Statistical learning theory , 1998 .

[4] Thorsten Joachims,et al. Making large scale SVM learning practical , 1998 .

[5] Peter Sollich. Probabilistic interpretations and Bayesian methods for support vector machines , 1999 .

[6] Matthias W. Seeger,et al. Covariance Kernels from Bayesian Generative Models , 2001, NIPS.

[7] Tom Minka,et al. A family of algorithms for approximate Bayesian inference , 2001 .

[8] Bernhard Schölkopf,et al. Estimating a Kernel Fisher Discriminant in the Presence of Label Noise , 2001, ICML.

[9] Bernhard Schölkopf,et al. Cluster Kernels for Semi-Supervised Learning , 2002, NIPS.

[10] L. Csató. Gaussian processes:iterative sparse approximations , 2002 .

[11] Neil D. Lawrence,et al. Fast Sparse Gaussian Process Methods: The Informative Vector Machine , 2002, NIPS.