Bidirectional Memristive Associative Memory for Adaptive Human–Machine Interaction in Emotion-Aware IoT Networks

Abstract

Association is not a singular associative process, different emotional states affect associative responses. Most memristor-based multilateral associative circuits aim to optimize associative patterns, neglecting the combination of emotional states and bidirectional association. Based on the emotional association mechanism of brain, a memristor-based circuit for two-dimensional emotional recall and bidirectional associative relationships is proposed. The circuit model continuously modifies the storage and recall processes of emotional memory, persistently adjusting two-dimensional emotions and revealing their impact on bidirectional association and self-regulation. The circuit consists of the temporal lobe cortex module, amygdala module, hippocampus module, prefrontal cortex module and hypothalamus module. Emotional recall is achieved through the temporal lobe cortex module, amygdala module and hippocampus module. Two-dimensional emotions and emotional regulation are realized by the prefrontal cortex module. Emotional state responses are realized by the hypothalamus module. Ultimately, this circuit is applied to machine fault prediction, providing a new reference for bionic intelligent robot in the application of Internet of Things (IoT).