Frenetic Steering in a Nonequilibrium Graph

In traditional recognition tasks of neural networks a potential landscape or cost function guides the system towards patterns using gradient dynamics. That is not how brain works as its dynamics far from equilibrium. We present an alternative and proof principle for pattern recovery in nonequilibrium model whereby only time-symmetric kinetics are altered. As mathematical model, random walker on randomly-oriented complete graph subject to finite driving direction arcs. Some vertices represent patterns. A first algorithm constructs basins attraction these second updates factors transition rates, order quickly reach remain there sufficiently long time, whenever starting vertex basin attraction.