PPV Modeling of Memristor-Based Oscillators and Application to ONN Pattern Recognition

An efficient perturbation projection vector (PPV) modeling for a memristor-based oscillator is proposed and then applied to the simulation of the oscillatory neural network (ONN) for pattern recognition. Compared with the existing model, which is very time consuming, this novel modeling method features a fast and accurate simulation of a large-scale network with many oscillators. The proposed modeling is based on a highly efficient abstraction of the phase sensitivity characteristic of the memristor-based oscillator, i.e., its PPV, which enables reducing the complexity of the ONN description, speeding up the simulation considerably while retaining comparable precision.