A model of a physical neural network with on-the-fly learning capability developed at NovSU

Scientists have found a way to configure connections in neural networks without recompiling code