Hello Sir,
regarding to real life systems, I have a few questions that come along an example that I would like to approach:
Imagine we have a ball that starts at a certain point h_max and falls until it hits the ground h_0, then bounces up to h_max again and then falls again and so on.
Certainly, we can model this example with two states 1. "falling" and 2. "rising" with the inputs h_max and h_0.
Now, all the states and inputs I simply can define in a custom java class that represents the SUL and can be learned from. The "issue" is that I have to know how the system works internally.
I would like to simulate the scenario, where I don't know how the system works internally, i.e. I don't know about the states and conditions. I only get the height of the ball and the direction it is moving in per time unit and I know about the starting point of the ball h_max.
So it should be some sort of black-box system that serves as SUL.
How could I possibly set up such an SUL and learn only from the corresponding outputs of certain inputs?
All I currently know is, that I can create a learning setup in which the learner learns about the automaton, which is either a DFA or a Java Class that I have to create manually. I don't have information about learning the automaton from an SUL that only gives you outputs for certain inputs and how to set this up.
Is there any source that is available?
Thanks in advance.
Kind regards,
John