A CNN dynamic reconfigurable associative memory circuit

Abstract

The invention discloses an associative memory circuit with a dynamic reconfigurable CNN. The circuit comprises a first cell neuron circuit and a second cell neuron circuit which are coupled, wherein the first cell neuron circuit and the second cell neuron circuit are identical in circuit structure. Each of the first cell neuron circuit and the second cell neuron circuit comprises an external signal input end, a template parameter control end, a memory signal output end and an operational circuit, wherein the operational circuit is connected with the external signal input end, the template parameter control end and the memory signal output end and used for performing operational operation on external input signals, template A parameters, template B parameters and offset parameters to obtain an associative memory result. Through the circuit, dynamic switching among multiple memory templates can be realized, the structure is simple, a super-large-scale circuit can be easily realized, and stability is high.

Description

technical field [0001] The invention belongs to the technical field of associative memory, and in particular relates to a CNN dynamic reconfigurable associative memory circuit. Background technique [0002] Associative memory is a method of using the connection between things to remember through association. In the neighborhood of neural computing, it can be described as a mechanism that simulates the information storage and recall of the human brain, enabling retrieval probes that meet certain conditions to find a series of standard templates pre-stored in the system. In traditional artificial neural network circuits, such as Hopfield network, the structure of neurons and their synapses is complex, which is not convenient for the realization of VLSI. [0003] Cellular neural network (CNN) is a large-scale nonlinear analog circuit that can process signals in parallel at high speed in real time, and is easy to implement in very large-scale circuits (VLSI). The logical behav...

AI Technical Summary

Problems solved by technology

In the traditional artificial neural network circuit, such as the Hopfield network, the construction structure of neurons and their synapses is complex, which is not convenient for the realization of VLSI

Images