Is it possible to train a neural network with one input, three outputs and one hidden layer?

Question

Neural networks

Data comes to the input, for example, within the range of 0 - 1000.
Can a neuron of three layers learn to respond to their change?

Answer 1

Can you clarify what you mean by "react to change"?
The fact is that the network must be trained - of course you understand this. So, in order to teach - you need to give examples. What are you giving as an example of change?
The second interpretation of your question lies in the field of clustering and searching for outliers, i.e. objects that are significantly different from those previously known. Here their methods are used.
In general - a description of your task in the studio and then perhaps we can help you with something.