Under what parameters are SVC and LinearSVC in scikit-learn equivalent?

by

In mathematical sense you need to set:

SVC(kernel="linear", **kwargs) # by default it uses RBF kernel

and

LinearSVC(loss="hinge", **kwargs) # by default it uses squared hinge loss

Another element, which cannot be easily fixed is increasing intercept_scaling in LinearSVC, as in this implementation bias is regularized (which is not true in SVC nor should be true in SVM – thus this is not SVM) – consequently they will never be exactly equal (unless bias=0 for your problem), as they assume two different models

  • SVC : 1/2||w||^2 + C SUM xi_i
  • LinearSVC: 1/2||[w b]||^2 + C SUM xi_i

Personally I consider LinearSVC one of the mistakes of sklearn developers – this class is simply not a linear SVM.

After increasing intercept scaling (to 10.0)

SVMs

However, if you scale it up too much – it will also fail, as now tolerance and number of iterations are crucial.

To sum up: LinearSVC is not linear SVM, do not use it if do not have to.

More Related Contents:

  1. How should I teach machine learning algorithm using data with big disproportion of classes? (SVM)
  2. Is this possible to predict the lottery numbers (not the most accurate)? [closed]
  3. How to interpret MSE in Keras Regressor
  4. Getting a low ROC AUC score but a high accuracy
  5. Why does one hot encoding improve machine learning performance? [closed]
  6. Multi-class classification in libsvm [closed]
  7. Retraining after Cross Validation with libsvm
  8. How to put more weight on certain features in machine learning?
  9. using precomputed kernels with libsvm
  10. 10 fold cross-validation in one-against-all SVM (using LibSVM)
  11. Early stopping with Keras and sklearn GridSearchCV cross-validation
  12. Scikits-Learn RandomForrest trained on 64bit python wont open on 32bit python
  13. Scikit learn – fit_transform on the test set
  14. Does the SVM in sklearn support incremental (online) learning?
  15. Determining the most contributing features for SVM classifier in sklearn
  16. Which algorithms involves in deep learning? [closed]
  17. Whats the difference between a test set and a train set?
  18. What is the role of the bias in neural networks? [closed]
  19. Is it possible to specify your own distance function using scikit-learn K-Means Clustering?
  20. Many to one and many to many LSTM examples in Keras
  21. How does sklearn.svm.svc’s function predict_proba() work internally?
  22. Caffe | solver.prototxt values setting strategy
  23. LabelEncoder for categorical features?
  24. What is the difference between loss function and metric in Keras? [duplicate]
  25. How to interpret loss and accuracy for a machine learning model [closed]
  26. What is the difference between linear regression and logistic regression? [closed]
  27. Keras: model.evaluate vs model.predict accuracy difference in multi-class NLP task
  28. How to compute precision, recall, accuracy and f1-score for the multiclass case with scikit learn?
  29. scikit-learn: how to scale back the ‘y’ predicted result
  30. Using Smote with Gridsearchcv in Scikit-learn

Leave a Comment