Learning an attractor network to recognize patterns.

<p><a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0002577#pone-0002577-g001" target="_blank">Figure 1</a>, top, (left): a network with 100 units with connections of random strength and with no specific pattern learned (for clarity only neighbouring connections are shown). <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0002577#pone-0002577-g001" target="_blank">Figure 1</a>, top (middle and right): learning pattern ‘A’. Top (middle): presenting ‘A’; top (right): connections between units related to ‘A’ are strengthened. <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0002577#pone-0002577-g001" target="_blank">Figure 1</a>, bottom, (left): presenting part of ‘A’. Bottom (right): the network has succeeded in retrieving ‘A’