Abstract:As the limits of transistor technology are approached, feature size in integrated circuit transistors has been reduced very near to the minimum physically-realizable channel length, and it has become increasingly difficult to meet expectations outlined by Moore's law. As one of the most promising devices to replace transistors, memristors have many excellent properties that can be leveraged to develop new types of neural and non-von Neumann computing systems, which are expected to revolutionize information-processing technology. This survey provides a comparative overview of research progress on memristors. Different memristor synaptic devices are classified according to stimulation patterns and the working mechanisms of these various synaptic devices are analyzed in detail. Crossbar-based memristors have demonstrated advantages in physically executing vector-matrix multiplication and enabling highly power-efficient and area-efficient neuromorphic system designs. The extensive uses of crossbar-based memristors cover in-memory logic, vector-matrix multiplication, and many other fundamental computing operations. Furthermore, memristor-based architectures for efficient neural network training and inference have been studied. However, memristors have non-ideal properties due to programming inaccuracies and device imperfections from fabrication, which lead to error or mismatch in computed results. To build reliable memristor-based designs, circuit-level, algorithm-level, and system-level solutions to memristor reliability issues are being studied. To this end, state-of-the-art realizations of memristor crossbars, crossbar-based designs, and peripheral circuitry are presented, which show both promising full-system inference accuracy and excellent power efficiency in multiple tasks. Memristor in-situ learning benefits from high energy efficiency and biologically-imitative characteristics, which are conducive to further realizing hardware acceleration of-cognitive learning. At present, the learning and training processes of brain-like networks are complex, presenting great challenges for network design and implementation.

High Robustness Memristor Neural State Machines

Reliable Memristor-based Neuromorphic Design Using Variation- and Defect-Aware Training

Memristors with Tunable Characteristics Based on the Tellurene/nb-Doped MoS2 Heterojunction Toward Bio-Realistic Synaptic Emulation

Memristor Neural Network Design

Essential Characteristics of Memristors for Neuromorphic Computing

A Memristor Neural Network Based on Simple Logarithmic-Sigmoidal Transfer Function with MOS Transistors

Design of Artificial Neurons of Memristive Neuromorphic Networks Based on Biological Neural Dynamics and Structures

Security enhancement of artificial neural network using physically transient form of heterogeneous memristors with tunable resistive switching behaviors

Leveraging Stochastic Memristor Devices in Neuromorphic Hardware Systems.

Build Reliable And Efficient Neuromorphic Design With Memristor Technology

Passive LiNbO₃ Memristor With Multilevel States for Neuromorphic Computing

A Flexible Memristor Model With Electronic Resistive Switching Memory Behavior and Its Application in Spiking Neural Network

Research Progress on Memristor: From Synapses to Computing Systems

Versatile memristor for memory and neuromorphic computing

Controlled Memory and Threshold Switching Behaviors in a Heterogeneous Memristor for Neuromorphic Computing

Memristor-based Neuromorphic Implementations for Artificial Neural Networks

Overview of Memristor-Based Neural Network Design and Applications

A Closed-Loop Design to Enhance Weight Stability of Memristor Based Neural Network Chips

MNSIM: Simulation Platform for Memristor-Based Neuromorphic Computing System.

Hardware implementation of memristor-based artificial neural networks

Dual functionality of NbOx memristors for synaptic and neuronal emulations in advanced neuromorphic systems