Skip to main content
. 2020 Jan 10;38(10):2349–2358. doi: 10.1007/s00345-019-03059-0

Fig. 1.

Fig. 1

Multilayer perceptron with input and output layer as well as an additional hidden unit layer. The neurons of a layer are represented as unfilled blue circles, and the completely colored neurons are the so-called bias neurons, whose value is constant and does not depend on the activations in previous layers. The weights of the compounds are learned in training (Source: [1])