Table shrinking bug in non-deterministic learning

[emuskardin]

Onfsm learning still can run into infinite loops, which should not happen with table shrinking! The idea behind is that the table is constantly updated with respect to all seen observations and this avoids infinite closing loops.

Smallest reproducible example

from Examples import onfsm_mealy_paper_example
from random import seed
seed(14)
onfsm_mealy_paper_example()

What happens is the infinite closing loop. Most likely cause is that you do not update properly the cells based on previously seen data.

Originally posted by @emuskardin in #29 (comment)

[emuskardin]

Made substantial progress on branch non-det-dev.

Things to note: all cells are resampled during learning, and it seems that setting lower n_resample with this technique leads to less total sampling. We should consider adding a tree and smarter sampling limit.

There is still a bug with such low sampling -> in rare cases even tho table is closed, extended row will not find a representative. This is weird bug. Even assert get_rows_to_close() is not Null did not help...

[emuskardin]

There is still a bug with such low sampling -> in rare cases even tho table is closed, extended row will not find a representative. This is weird bug. Even assert get_rows_to_close() is not Null did not help...

Fixed by computing whole extended S every time it is needed.

[emuskardin]

Solved on branch non-det-dev.
Branch will be merged once abstracted version is adapted.