Average classification performance.

<p>100 networks are trained by STDP and IP simultaneously (orange), IP alone (blue), STDP alone (green), or are nonplastic (gray). Optimal linear classifiers are then trained to perform (A) the memory task RAND x 4, (B) the prediction task Markov-85, and (C) the nonlinear task Parity-3. Nonplastic n... View More