Suppose we are fitting a LogisticRegression model with scikit-learn, or the same model with pytorch. In multiclass problems, the strategy OneVsRest will fit a different classifier for each of the labels of the dataset. For each classifier, the class will be fitted against all the other classes. I want to ask: do you know if there is some research paper or some evidence of a relationship among the different classifiers? I mean, each classifier will have its own weights. Will modifying the vector of weights of one classifier also impact the performance of the other classifiers?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.