Synthesizing a Neuron Using Chemical Reactions

Deep learning and synthetic biology are two highly popular fields and draw lots of attentions. Deep learning has been employed for complex problems, and researchers have developed synthetic biology into a powerful tool for More than Moore. However, few works have considered implementing deep neural networks (DNNs) with synthetic biology. In this paper, by revealing the common probability base, we aim to implement the most fundamental element of DNN, a neuron, using chemical reaction networks (CRNs). We firstly propose a computation model in CRNs, then present our architecture of a neuron using such computation model as a basis. The correctness of such computation model and architecture is proved by both mathematical derivation and silico simulation.