Abstract:
Oscillatory Neural Networks (ONN) are inevitable when it comes to solving combinatorial optimization problems. This work demonstrates the fault tolerance of the ONN in solving vertex coloring problems in a 4-node network in various configurations at multiple failure levels of the different components of the oscillator. This work validates the network to be extremely robust to failures (limited to 4 nodes), showing tolerance in variations in resistance and capacitances up to 90% and 50% respectively.