Skip to main content
. 2012 Mar 22;8(3):e1002430. doi: 10.1371/journal.pcbi.1002430

Figure 4. Feedforward neural networks: Evolutionary dynamics for two representative simulations, where.

Figure 4

Inline graphic and Inline graphic . The same graphic conventions as in Figure 2 are followed. (A) Inline graphic. Top graphs: Inline graphic increases to the optimal value 0.75; specialization remains low. Bottom graphs: connection weights incoming at output node 1 become positive (strongest connection weight being Inline graphic); connection weights incoming at output node 2 become negative (Inline graphic) or positive, but very close to zero (Inline graphic). (B) Inline graphic. Top graphs:Inline graphic increases to values above 0.75; Inline graphic remains low. Bottom graphs: similar to when Inline graphic, but Inline graphic is closer to zero.