Memristor-Based Neural Network Circuit of Memory With Emotional Homeostasis
Junwei Sun, Juntao Han, Yanfeng Wang
Abstract
Most artificial neural networks based on memristor only consider memory and ignore other factors that may affect memory, such as emotion. In this paper, a memristive neural network circuit that can realize memory with emotional homeostasis is designed. It mainly contains three modules: an auditory module, an emotion module and a memory module. The auditory module mimics the auditory receiver in the brain, which can receive external music stimuli and generate auditory voltage. The emotion module realizes a conversion mechanism from hearings to emotions. The extent of emotion will increase with the increase of auditory voltage. In addition, the function of emotional homeostasis is also implemented by the emotion module. If the auditory voltage exceeds a certain value, the extent of emotion will not increase but decrease. The memory module implements the function of learning and forgetting based on the biological synapse composed of memristors. The simulation results in PSPICE show that the proposed circuit can learn and forget under various types of music like humans. The memory neural network with emotional generation and homeostasis provides a reference for further development of brain-like systems.