They tried to technique the trouble with several symbolic methods, together with what were then termed "neural networks"; these had been mostly perceptrons together with other styles which were later found to become reinventions from the generalised linear models of data.[twenty] Probabilistic reasoning was also utilized, especially in automated me