Hands-On Artificial Intelligence for Cybersecurity

SVM spam filter example

Let's go back to our sample spam filter, and replace the Perceptron with an SVM, as we have seen in the identification of the hyperplane that we are not limited to using linear classifier models only (being able to choose between classifiers characterized by greater complexity).

However, to compare the results obtained previously with the Perceptron, which represents a strictly linear classifier, we will also choose a linear classifier in the case of the SVM.

This time, however, our dataset (stored in the sms_spam_svm.csv file, and derived from the collection of SMS spam messages we found earlier in the chapter, in which the total occurrences of the various suspicious keywords were extracted and compared to the total ...