Is it correct not to use the threshold function on the last layer of the neural network?

Question

Neural networks

I want to build a neural network, where the input will be an array of numbers and the output will also be an array of numbers (not boolean). In the primitive case, this is just a fit, but if the network is multilayer, then it is desirable to use the logistic function. So far I have not seen examples of such use of ns.

Answer 1

It will be quite correct. if the values ​​at the input and output are normalized, then at the outputs you will most likely receive a probability value (in the case of classification / clustering problems)