In the artificial intelligence (AI) era, highly intelligent tasks such as real-time big data analysis, speech/face recognition, self-driving automobile navigation, and even surpassing human capabilities became feasible. These tasks, involving the processing of large amounts of unstructured information, rely on graphic processing units-based huge server systems to train AI algorithms. However, these AI systems face significant challenges due to high power consumption and limited computing performance. Particulary, the increasing trend of deep neural network parameters causes huge power consumption and large area overhead of a nonlinear neuron electronic circuit, and it incurs a vanishing gradient problem. To overcome these challenges, research is underway to replace the von Neumann architecture with hardware-based artificial neural networks (ANNs) known as brain-inspired neuromorphic systems. To achieve complete energy-efficient neuromorphic systems, it is also important to research not only synaptic device of memristor crossbar array but implementation of neuron processors to compute and propagate activation.