Target function and features

[Merce V]

Taylor explains that the target function can be formulated in different ways, but limits his explanation to a function that returns a distance. However, he mentions that it could also be formulated as a generative probabilistic model. Is there a significant difference between using one or the other approach?

I would also like to know a bit more about how the distance is calculated for different types of features, for instance, binary-valued, continuous or N-ary categorical features. Can we calculate the Euclidean distance for all of them? Or do we need to somehow manipulate them beforehand?

[Simon]

We’ll cover both points in the lecture.

[Author]

Posts