Design And Implementation Of Memristor-based Associative Memory Neuromorphic Circuit

In the 1940 s and 1950 s,the continuous development of atomic energy,computer technology,space technology and other fields promoted the arrival of the third scientific and technological revolution,which brought great changes to human production and life.Now,standing at the crossroads of the fourth scientific and technological revolution,countries around the world are also gearing up to join the future science and technology competition.Artificial intelligence is predicted by scientists to be one of the protagonists of the fourth scientific and technological revolution.In this context,memristor neuromorphic circuits are considered to be the most promising platform for research on brain-like intelligence due to their advantages in integration and mimics.In recent years,many scholars have made important contributions to the development and research of brain-like intelligence by designing synaptic structural circuits and neuron structural circuits and constructing neuromorphic circuits to simulate the biological mechanism of brain.At present,the function of analog associative memory is too simple,and most of them use computers or digital chips to assist the calculation,which is not conducive to the realization of hardware circuit imitation and large-scale integration design.Therefore,the neuromorphic circuit with more biological characteristics is studied in this paper,and no other devices are used to assist calculation.It solves the deficiency in the existing research.The details are as follows:(1)Based on the study of Pavlov’s biological experiments,this paper proposes a new neuron circuit based on IF neuron.It not only has a simple circuit structure,but also can eliminate the interference of instantaneous pulse and avoid the propagation and flooding of burrs in the circuit.Based on the abstract experimental model,the synaptic circuit is constructed.The synapse circuit and neuron circuit are combined to form the memristor associative memory circuit.The successful realization of learning,accelerated learning,association,forgetting,decelerating forgetting,decelerating natural forgetting and other functions,and the learning and forgetting rate can be changed many times,making the circuit more biogenic.We also combine the circuit with the Ebbinghaus forgetting experiment,and successfully simulate the Ebbinghaus forgetting rule by changing the input of the circuit.This shows that the circuit has better applicability and expands the application range of the circuit.It provides ideas and methods for constructing more comprehensive and complex brain-like circuits.(2)There are a variety of receptors or sensory tissues in the human body,which can receive external information and transmit it to the brain through the nerve center for information communication and exchange,and produce associative memory and physiological stress action.Based on this biological mechanism,we construct an information transmission model: when human ingests food,the taste system and the visual system work simultaneously.Different foods have different appearance characteristics and tastes,which form associative memory in the central nervous system,and different flavors stimulate the face to make different stress expressions.The visual-gustation associative memory circuit is constructed by this model.Different food stimuli make facial neurons express different facial expressions,successfully realizing associative memory between visual and taste systems.The circuit completely simulates the brain neural action process of food intake,which provides a reference for the realization of higher level artificial intelligence system.