You are given a dataset with 500 examples, 20 features and 2 classes. the data is linearly separable, and thus, you train a linear svm using 5-fold cross validation obtaining the following results: fold 1 fold 2 fold 3 fold 4 fold 5 error 1/100 36/100 1/100 0/100 1/100 the classifier is performing very well except the 2nd fold. can you give a plausible explanation? (hint: think about the support vectors)