It is well-known that deep feedforward networks can approximate any continuous function from $\mathbb{R}^k$ to $\mathbb{R}^l$, (uniformly on compacts).

However, in practice feature maps are typically used to improve the learning quality and likewise, readout maps are used to make neural networks suited for specific learning tasks.

For example:

  • Classification: networks are composed with the softmax (readout) function so they take values in $(0,1)^l$.

What are examples of commonly used feature and readout maps?

  • 1
    $\begingroup$ @nbro I have reformulated the question to reduce generality and clarify what I'm looking for... (letting you know in advance) $\endgroup$
    – ABIM
    Apr 1, 2020 at 7:32
  • $\begingroup$ You gotta be kidding me! This is a completely different question! Now, you're not even talking about graph neural networks. Please, stop editing your posts every time you have a new question. If you have a new question, open a new thread! $\endgroup$
    – nbro
    Apr 1, 2020 at 12:25


You must log in to answer this question.