4.4. Linear SVMs: Fuzzy Separation
Sometimes the training data points are not clearly separable, and most often, they are best characterized as "fuzzily" separable. It is therefore convenient to introduce the notion of a fuzzy (or soft) decision region to cope with such situations. For the nonseparable cases, it is not possible to construct a linear hyperplane decision boundary without incurring classification errors. A fuzzy SVM is a model that allows a more relaxed separation, which in turn would facilitate a more robust decision.
For fuzzy SVM classifiers, it is no longer suitable to talk about the separation width. Instead, it is more appropriate to adopt a notion of fuzzy region or, more exactly, fuzzy separation region. Accordingly, the ...
Get Biometric Authentication: A Machine Learning Approach now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
