Neruomorphic Circuit Implementation for Bio-Inspired Self-Regulating Mutual Associative Memory Network

Associative memory is one of the necessary conditions for advanced organisms to have cognitive ability and for in-telligences to acquire knowledge and experience from the environment. In this paper, a self-regulating mutual associative memory network circuit is constructed based on biological mechanisms and a novel memory device, the memristor. The circuit includes a pulse width delay (PWD) module, a mutual associative learning-retrieval module and a self-regulation module. The PWD module en-ables the formation of associative memory between non-simulta-neous input features; the mutual associative learning-retrieval module enables bidirectional association and retrieval between features; the self-regulation module enables the associative learning rate to change according to the duration of consolidation learning, and is also the key module for the conversion of short-to-long-term memory. The simulation verified the process of enhancement and weakening of associative memory, the change of associative memory rate with input stimulus interval, and the process of short-to-long-term memory conversion. The circuit has the ad-vantages of multifunctionality, low area, and low power consumption, and is expected to facilitate the development of embodied in-telligence that realizes advanced cognitive functions.