Week 4 Flashcards
What is the idea of SVM?
What is the hyperplane that SVM chooses?
What is the margin in SVM, and support vectors?
What is the technical definition of SVMs?
What is the definition of a hard margin SVM?
Why is hard margin SVM not ideal?
What is soft margin SVM? What is its definition?
What are common values for C in the soft-margin SVM? When is the hard margin SVM recovered?
What are the two options for solving soft margin SVM problems?
How to rewrite the soft-margin SVM as a unconstrained optimization problem?
How to solve Soft margin SVM directly?
What is the idea behind extending the feature space?
What is the direct SVM problem with a polynomial kernel?
What is the Radial Basis Function kernel?
Compare SVM vs NN.
How does Support Vector Regression work?
What should the requirements of an ML project specify?
What are the two options of dealing with categorical data
How to deal with missing values?
How to enforce normilization?
What is data augmentation? What are some challenges?
What is the idea of model learning?
What is the difference between hyperparameters and internal parameters?
What is the train/dev/test split?
What is the size of the train/dev/test groups?
What is the k-fold Cross-validation?
What is the generalization ability?
When is accuracy not a good indicator of performance?
What is a confusion matrix?
Which metrics can be derived from a confusion matrix?
