Suppose I trained a Gaussian process classifier with a linear kernel (using GPML toolbox) and got some feature weights for each input feature.

My question is then:

Does it/When does it make sense to interpret the weights to indicate the real-life importance of each feature or interpret at group level the average over the weights of a group of features?

  • $\begingroup$ Hi Alex and welcome to this community! I suggest you ask only one question. I already doubt that there's anyone here that has a better knowledge of Gaussian processes than me, so, if you want to attract the attention of people to answer your question (by facilitating the reader's life), then simplify your post and ask just one question. Ask the other questions in separate posts (each in its own post). $\endgroup$
    – nbro
    Dec 1 '19 at 2:31
  • 1
    $\begingroup$ Thank you for the advice, @nbro ! :) I have edited the question accordingly. Also if you have any insights I would be very grateful to be enlightened! ;) $\endgroup$
    – Alex
    Dec 1 '19 at 11:18
  • $\begingroup$ How can you get the weights? The weights are given as an input, aren't they? I assume by "weight" you mean this part of input which is used to calculate weight of each input in the likelihood? Or what exactly you mean by "weight"? $\endgroup$
    – Tomas
    Jan 13 at 10:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.