Comparative analysis of memristor devices as neuron

Abstract

Memristor, a two-terminal device whose unique ability to control its resistance by varying its input, is a promising technological device for intelligent systems. In humans, pattern recognition is associated with memory which helps in decision making for the behavioral and cognitive aspects. The storage of memory is crucial in decision making, and the adaptation between remembering and forgetting memory in time plays a pivotal role for pattern decision; this memory consolidation method is an extreme design challenge in the neuron, the detailed changes in the ion exchange and its timings is what responsible for memory consolidation. A memristor is recognized as a device able to mimic and function like a synapse. Recent advances in memristor have shown the working of a memristor as the different levels of memory, i.e. Long-term memory (LTM) and Short-term memory (STM), memristor devices perform memory potentiation and depression characteristics in them. However, to perform as a real neuron, a network of multiple memristors with careful design has to be followed. In this paper, the detailed mechanism of neuron morphology is explained to show how the many mechanisms inside the biological neuron become responsible for memory consolidation and compare how a memristor can function the neuron rules like plasticity and spike-threshold.