Information Geometrically Generalized Covariate Shift Adaptation

Many machine learning methods assume that the training and test data follow the same distribution. However, in the real world, this assumption is often violated. In particular, the marginal distribution of the data changes, called covariate shift, is one of the most important research topics in machine learning. We show that the well-known family of covariate shift adaptation methods is unified in the framework of information geometry. Furthermore, we show that parameter search for a geometrically generalized covariate shift adaptation method can be achieved efficiently. Numerical experiments show that our generalization can achieve better performance than the existing methods it encompasses.

Automated summary

Many machine learning methods assume the same distribution for training and test data, but this assumption is often violated in the real world. Covariate shift, the change in marginal distribution of data, is a significant research topic in machine learning. We have shown that the well-known family of covariate shift adaptation methods can be unified within the framework of information geometry. Additionally, we have demonstrated that parameter search for geometrically generalized covariate shift adaptation method can be efficiently achieved. Numerical experiments indicate that our generalization outperforms existing methods it encompasses.