Academic
Publications
Comment on "On Discriminative vs. Generative Classifiers: A Comparison of Logistic Regression and Naive Bayes

Comment on "On Discriminative vs. Generative Classifiers: A Comparison of Logistic Regression and Naive Bayes,10.1007/s11063-008-9088-7,Neural Process

Comment on "On Discriminative vs. Generative Classifiers: A Comparison of Logistic Regression and Naive Bayes   (Citations: 2)
BibTex | RIS | RefWorks Download
Comparison of generative and discriminative classiflers is an ever-lasting topic. Based on their theoretical and empirical comparisons between the na˜‡ve Bayes clas- sifler and linear logistic regression, Ng and Jordan (2001) claimed that there existed two distinct regimes of performance between the generative and discriminative clas- siflers with regard to the training-set size. However, our empirical and simulation studies, as presented in this paper, suggest that it is not so reliable to claim such an existence of the two distinct regimes. In addition, for real world datasets, so far there is no theoretically correct, general criterion for choosing between the discrim- inative and the generative approaches to classiflcation of an observation x into a class y; the choice depends on the relative confldence you have in the correctness of the speciflcation of either p(yjx) or p(x;y). This can be to some extent a demonstra- tion of why Efron (1975) and O'Neill (1980) prefer LDA but other empirical studies may prefer linear logistic regression instead. Furthermore, we suggest that pairing of either LDA assuming a common diagonal covariance matrix (LDA-⁄) or the na˜‡ve Bayes classifler and linear logistic regression may not be perfect, and hence it may not be reliable for any claim that was derived from the comparison between LDA-⁄
Journal: Neural Processing Letters - NPL , vol. 28, no. 3, pp. 169-187, 2008
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
Sort by: