The idea is simple.
Say you have different types of processes in a cycle. If you assign each process a number then the cycle will generate a curve or a pattern. What I wonder is if one could use time prediction or predict inbetween processes by the patterns from other physically real events.
With this you could have something that looks “process realistic”.
An example would be to assign a number to reversible and irreversible type of processes.
My speculation is that with this. You could reverse guess a needed object. From the process number that the machine learning algorithm has predicted.