What is Support Vector Machine (SVM)
Support Vector Machine (support vector machine)
Differences between Support Vector Machine and Perceptron
Assume you have two categories of data points (such as circles and crosses):
- Perceptron: Just find a line that can separate the circles and crosses, any line will do.
- SVM: Not only separate them, but also make this line as far away from all points as possible (maximum margin), so that even if new data slightly shifts, it is less likely to be misclassified.
You can think of SVM as an “upgraded version” of the perceptron: Perceptron → Add hinge loss to become the margin perceptron → Add regularization to become the Support Vector Machine (SVM)