Question

Suppose you’ve trained an SVM classifier with an RBF kernel, but it seems to underfit the training set. Should you increase or decrease γ (gamma)? What about C? Explain your reasoning. Train an SVM classifier on the MNIST dataset. Since SVM classifiers are binary classifiers, you will need to use one-versus-the-rest to classify all 10 digits. You may want to tune the hyperparameters using small validation sets to speed up the process. What accuracy can you reach?

Name: suppose youve trained an svm classifier with an rbf kernel but it seems to underfit the training set should you increase or decrease gamma what about c explain your reasoning train an svm cl 86841
Uploaded: 2022-06-25T16:08:41-08:00
Duration: 5 min 39 s
Channel: Shyam P
Description: suppose youve trained an svm classifier with an rbf kernel but it seems to underfit the training set should you increase or decrease gamma what about c explain your reasoning train an svm cl 86841

          Suppose you’ve trained an SVM classifier with an RBF kernel, but it seems to underfit the training set. Should you increase or decrease γ (gamma)? What about C? Explain your reasoning. Train an SVM classifier on the MNIST dataset. Since SVM classifiers are binary classifiers, you will need to use one-versus-the-rest to classify all 10 digits. You may want to tune the hyperparameters using small validation sets to speed up the process. What accuracy can you reach?

Added by Lauren G.

Computer Science and Information Technology

Trishna Knowledge Systems 2018 Edition

Instant Answer

Solved by Expert Shyam P

06/25/2022

Step 1

- The RBF kernel has two important hyperparameters: γ (gamma) and C. - γ controls the influence of a single training example. A small γ means a far reach (smooth decision boundary), while a large γ means a close reach (complex decision boundary). - C controls the Show more…

Show all steps

Thanks for your feedback!

Suppose you’ve trained an SVM classifier with an RBF kernel, but it seems to underfit the training set. Should you increase or decrease γ (gamma)? What about C? Explain your reasoning. Train an SVM classifier on the MNIST dataset. Since SVM classifiers are binary classifiers, you will need to use one-versus-the-rest to classify all 10 digits. You may want to tune the hyperparameters using small validation sets to speed up the process. What accuracy can you reach?