The Hopfield Net and Boltzmann Machine
The MIT Press eBooks · 2018
Computer scientist Jerome Feldman was at the University of Rochester when he embraced a connectionist network approach to artificial intelligence in the 1980s. Ever the truth teller, Jerry pointed out that the algorithms being used in AI took billions of steps to reach an often incorrect conclusion, whereas the brain could reach a usually correct conclusion in around 100 steps. 1 Feldman's “100-step rule” was not popular among AI researchers at the time, but a few, most notably Allen Newell at Carnegie Mellon, did use it as a constraint.