#support-vectors

Question 1: Select all that apply

Which statements about $k$ -NN and SVMs with an RBF kernel are true? Select all that apply.

$k$ -NN may perform poorly in high-dimensional spaces (say, $d > 1000$ )1
In an RBF SVM, large values of gamma tend to give a higher training score but probably a lower validation score2
If we increase both gamma and C, we cannot tell whether the model becomes more or less complex3
Like $k$ -NN, an RBF SVM must keep every training example to make predictions4