Support Vector Machines

Question 1

what are support vector machines?

Answer 1

a model capable of performing linear and nonlinear classification, regression and even outlier detection.

Question 2

what sized datasets are SVMs best suited for?

Answer 2

small or medium sized ones

Question 3

what can you think of an SVM classifier as?

Answer 3

as fitting the widest possible street between classes (largest margin classification)

Question 4

What are downfalls of SVMs?

Answer 4

Sensitive to features scales

Question 5

what is soft margin classification?

Answer 5

finds a good balance between keeping the street as large as possible and limiting the margin violations

Question 6

What to do if your SVM is overfitting?

Answer 6

regularize it by reducing soft margin parameter C

Question 7

What does Logistic Regression classifiers do that SVMs dont?

Answer 7

output probabilities

Question 8

what is a nonlinear svm classifier?

Answer 8

when we add more nonlinearly features (polynomial) when a svm set is non linearly separable

Question 9

whats bad about a high polynomial degree?

Answer 9

creates a huge number of features, making the model too slow

Question 10

whats bad about a low polynomial degree?

Answer 10

cannot deal with very complex datasets

Question 11

what is the kernel trick?

Answer 11

stops huge number of features being created