The New Turing Omnibus Errata

Chapter 36 - Neural Networks That Learn

p. 243

• The hyperbolic tangent function is actually

p. 247

• Line 8 of the back-propagation algorithm should be sigmoid(i) <- 1 - ((medout(i))^2
• The final line of the back-propagation algorithm should be synone(i, j) <- synone(i, j) + delta