How to train a neural network whose output will be the probability of an event occurring?

Question

Neural networks

How to train a neural network whose output will be the probability of an event occurring?
And is it possible in principle, if not, then advise another approach.
For example, there are two outcomes of an event (Outcome1 and Outcome2). There is a set of input data on the basis of which it is possible to predict the probability of these outcomes. Those. the result will be a probability of 0..1. Therefore, the occurrence of Exodus 2 will be correct, even if a negligible probability of its occurrence is predicted.

Answer 1

What is the problem?
At the output of the neural network, you will have some number for event 1 and some for event 2.
If these are some non-normalized numbers, then use softmax