What does this network response mean?

Question

Neural networks

There is a neural network that is used in Unity, it seems to be trained about Hebb. I need it to classify the situation and give 1 out of 3 actions for the situation.

In the FixedUpdate method, I pass the data to it for training, and each time I adjust the weights according to the Hebb rule.

At the output layer, there is 1 neuron, which, after several tens of iterations, starts outputting 0.017 and gradually decreases.

What conclusions can be drawn from this? How to classify situations?

Answer 1

In classification, the output dimension is usually equal to the number of classes.
1 out of 3 actions == 3 "neurons".
On each a certain value from 0 to 1 - the probability. Take the option with the highest probability.