Composition methods for the integration of dynamical neural networks

We apply the symmetric composition method for the integration of ordinary diierential equations to dynamical neural networks. In this method, we split the vector eld, which is parameterized by a neural network, into the contribution of each of its neurons. We then solve the elementary diierential equation associated to each neuron separately , and recombine these contributions in a sequence of compositions. This gives rise to simple integration rules for dynamical neural networks, which we present for dynamical single-hidden-layer perceptrons.