ABSTRACT

This chapter includes three closely related applications of principal component analysis (PCA). First, it briefs a technique known as eigenfaces, which is a PCA-based approach to facial recognition. Then it turns the attention to the problem of malware detection and shows that essentially the same approach is effective in this problem domain. Finally, it considers the use of PCA for the challenging problem of image spam detection. Facial recognition is an interesting problem in artificial intelligence, and a topic of considerable interest in security. In the security domain, facial recognition can be viewed as a type of biometric, and has been proposed, for example, as a means of detecting terrorists in airports. In eigenvector analysis is applied to the facial recognition problem, where it's given the clever name of eigenfaces. An analogous eigenviruses technique is applied to the malware detection problem in, and the same technique is further analyzed in.