They attempted to approach the issue with numerous symbolic procedures, and what were being then termed "neural networks"; these were being mainly perceptrons and various products that were afterwards uncovered to get reinventions from the generalised linear styles of figures.[twenty] Probabilistic reasoning was also used, particularly in automated health-related analysis.[21]: https://machine-learning53186.jiliblog.com/93238672/customer-loyalty-can-be-fun-for-anyone