Expandable Neuron Cell with Current-Steering DAC for Deep-Learning

For the past few years, AI and deep-learning got a second wind in the development of Brain neuroscience and Big data. In this paper, we describe an analog neuron model by using current-steering DAC circuit which uses analog circuits simulate the internal architecture of the neuron: Dendrites, axon, synapses and soma. The architecture we propose can achieve most of the network architecture, and the expandable neuron architecture that the hardware architecture can be adjusted according to the size of the system for better energy efficiency.