Skip to main content
. 2019 Mar;23(3):235–250. doi: 10.1016/j.tics.2018.12.005

Figure I.

Figure I

Artificial Neural Networks. (A) Layers of neuron-like nodes are represented by sets of stacked blue circles. Feedforward connections are indicated by green arrows. (B) Prediction. (C) Learning. (D) Schematic of the directions of two consecutive weight modifications (thick arrows) in the space of weights (for simplicity, only two dimensions are shown). Contours show points in weight space with equal cost function values.