[Andrs]

In lecture 18("Epilogue"), there is a nice map about machine learning (slide 4/23). One class is "Models" that describes the different algorithms used in machine learning. Some of the algorithms under models are "Linear", "SVM", "Neural Networks", "nearest neighbors", etc.
I have a question about the class of models that are "linear". The most basic linear algorithm is the perceptron. But I would say that SVM (Support Vector Machine)could also be defined as a linear algorithm as well as "Linear Discriminant Analysis" and "linear regression" (not well fitted for classif but it is a linear model). That is, under the linear umbrella, there are many different algorithms. Of course, things starts becoming more difuse when we use z-transform (or kernel) in the linrear algorithms. But still there is the question about the criteria and purpose to create a generic linear class. What was the intention to have a "Linear class" and a different class called "SVM" under the models?