The score used here is relative to the expected hardness of the task. It is obtained by dividing the score of the algorithm
(percentage of strings with good classification) by the score of the baseline.
Small automata
Medium automata
Large automata
Small alphabet
AL - 19 states - 94.11/71.67=1.31 Jure Žbontar - 18 states - 86.61/71.78=1.21
Jure Žbontar - 94 states - 93.11/74.11=1.26 AL - 81 states - 89.83/76.83=1.17
AL - 118 states - 83.39/70.94=1.18 Jure Žbontar - 115 states - 94.17/81.22=1.16
Medium alphabet
Jure Žbontar - 17 states - 99.17/76.83=1.29 manolo - 12 states - 93.89/76.83=1.22
manolo - 87 states - 92.83/72.94=1.27 Jure Žbontar - 45 states - 91.11/78.22=1.16