A very popular model in machine learning is the feedforward neural network (FFN). After a brief introduction to machine learning, we describe FFNs which represent sections of holomorphic line bundles on complex manifolds, and software which uses them to get numerical approximations to Ricci flat Kähler metrics.