Building Recurrent Neural Networks to Implement Multiple Attractor Dynamics Using the Gradient Descent Method

<table>Teaching sequences of experiment 2. (a) Trajectories on <svg height="15.975" id="M144" style="vertical-align:-0.1092pt" version="1.1" viewbox="0 0 19.049999 15.975" width="19.049999" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
<g transform="matrix(.017,-0,0,-.017,.062,15.775)"><path d="M751 18q-1 -4 -2 -13t-2 -13q-24 0 -35 1q-57 3 -92.5 23t-69.5 68q-21 30 -101 160q-13 22 -30.5 31t-54.5 9h-32v-159q0 -62 14 -77t74 -20v-28h-382v28q62 5 76 19.5t14 77.5v400q0 63 -14 77.5t-76 19.5v28h372q109 0 159 -34q67 -42 67 -131q0 -115 -127 -172
q27 -52 89 -144q49 -71 81 -104q31 -36 72 -47zM543 469q0 76 -42 112t-104 36q-41 0 -54 -10q-11 -7 -11 -44v-247h48q73 0 108 29q55 42 55 124zM290 131v390q0 62 -7.5 79.5t-30.5 17.5q-24 0 -31.5 -17.5t-7.5 -79.5v-390q0 -62 7.5 -79.5t31.5 -17.5q23 0 30.5 17.5
t7.5 79.5z" id="x211D"></path></g>
<g transform="matrix(.012,-0,0,-.012,12.663,7.613)"><path d="M412 140l28 -9q0 -2 -35 -131h-373v23q112 112 161 170q59 70 92 127t33 115q0 63 -31 98t-86 35q-75 0 -137 -93l-22 20l57 81q55 59 135 59q69 0 118.5 -46.5t49.5 -122.5q0 -62 -29.5 -114t-102.5 -130l-141 -149h186q42 0 58.5 10.5t38.5 56.5z" id="x32"></path></g>
</svg>.
(b) Temporal trajectories of teaching sequences.</table>

Advances in Artificial Neural Systems

fig6

Figure 6

Figure 6: Building Recurrent Neural Networks to Implement Multiple Attractor Dynamics Using the Gradient Descent Method 