Discriminative Learning Under Covariate Shift

We address classification problems for which the training instances are governed by an input distribution that is allowed to differ arbitrarily from the test distribution---problems also referred to as classification under covariate shift. We derive a solution that is purely discriminative: neither training nor test distribution are modeled explicitly. The problem of learning under covariate shift can be written as an integrated optimization problem. Instantiating the general optimization problem leads to a kernel logistic regression and an exponential model classifier for covariate shift. The optimization problem is convex under certain conditions; our findings also clarify the relationship to the known kernel mean matching procedure. We report on experiments on problems of spam filtering, text classification, and landmine detection.

[1]  Steven R. Lerman,et al.  The Estimation of Choice Probabilities from Choice Based Samples , 1977 .

[2]  J. Heckman.  Sample selection bias as a specification error , 1979 .

[3]  R. Pyke,et al.  Logistic disease incidence models and case-control studies , 1979 .

[4]  D. Rubin,et al.  The central role of the propensity score in observational studies for causal effects , 1983 .

[5]  P. J. Green,et al.  Density Estimation for Statistics and Data Analysis , 1987 .

[6]  Thorsten Joachims,et al.  A Probabilistic Analysis of the Rocchio Algorithm with TFIDF for Text Categorization , 1997, ICML.

[7]  H. Shimodaira,et al.  Improving predictive inference under covariate shift by weighting the log-likelihood function , 2000 .

[8]  Charles Elkan,et al.  The Foundations of Cost-Sensitive Learning , 2001, IJCAI.

[9]  JapkowiczNathalie,et al.  The class imbalance problem: A systematic study , 2002 .

[10]  Nathalie Japkowicz,et al.  The class imbalance problem: A systematic study , 2002, Intell. Data Anal..

[11]  J. Lunceford,et al.  Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study , 2004, Statistics in medicine.

[12]  Bianca Zadrozny,et al.  Learning and evaluating classifiers under sample selection bias , 2004, ICML.

[13]  Masashi Sugiyama,et al.  Input-dependent estimation of generalization error under covariate shift , 2005 .

[14]  Miroslav Dudík,et al.  Correcting sample selection bias in maximum entropy density estimation , 2005, NIPS.

[15]  Bernhard Schölkopf,et al.  Correcting Sample Selection Bias by Unlabeled Data , 2006, NIPS.

[16]  Steffen Bickel,et al.  Dirichlet-Enhanced Spam Filtering based on Biased Samples , 2006, NIPS.

[17]  Motoaki Kawanabe,et al.  Direct Importance Estimation with Model Selection and Its Application to Covariate Shift Adaptation , 2007, NIPS.

[18]  Lawrence Carin,et al.  Multi-Task Learning for Classification with Dirichlet Process Priors , 2007, J. Mach. Learn. Res..

[19]  Steffen Bickel,et al.  Discriminative learning for differing training and test distributions , 2007, ICML '07.

[20]  Mehryar Mohri,et al.  Sample Selection Bias Correction Theory , 2008, ALT.

[21]  Masashi Sugiyama,et al.  Direct Density Ratio Estimation for Large-scale Covariate Shift Adaptation , 2008, SDM.

[22]  Masashi Sugiyama,et al.  Direct Density Ratio Estimation for Large-scale Covariate Shift Adaptation , 2009, J. Inf. Process..