M
M
MaxBat2021-09-23 21:46:48
Neural networks
MaxBat, 2021-09-23 21:46:48

How to get the number at the output of the neural network if the activation functions bring the values ​​inside the network to [0;1]?

I am writing an autoencoder and the output should be numbers from -100 to 100. How is it possible to get these numbers if the activation functions result in values ​​within the network from zero to one. That is, even the output neuron, if it receives a value of 100, will give out a sigmoid, for example, weave. And instead of 100 I will get 0.99. normalize somehow or some other activation functions, use others, or not use them at all, but then the logic of the neuron collapses.

Thank you.

Answer the question

In order to leave comments, you need to log in

1 answer(s)
R
rPman, 2021-09-23
@rPman

x*200-100
how are you going to work with neural networks if you don't even understand the simplest linear operations on numbers? how do you for example bring the input numbers to -1..+1?

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question