Abstract:Memristors serve as electronic components with the ability to store charge and demonstrate resistive states, which are similar to the synaptic connections between biological neurons. For this reason, they have been widely applied in the research and development of neuromorphic computing models to achieve functions similar to synaptic transmission. As early as the 1960s, some studies suggested that some animals could enter a sleep state in which one hemisphere of the brain is in an asleep state while the other remains active, controlling essential physiological functions like breathing, swimming, and avoiding danger. This phenomenon is known as uni-hemispheric slow-wave sleep (USWS). On the other hand, the echo state network (ESN) is a kind of neural network that has a simple but high-efficiency structure and has garnered significant attention lately. However, there is limited research on using ESN and MEM to simulate USWS. In this paper, we propose a new echo state network model based on memristors, which is used to simulate the brain of animals with USWS. In our model, two ESNs based on memristors are used as the left and right brains to switch between sleep and wakefulness. In addition, we use an elastic network to optimize the output, which combines the advantages of ridge regression and lasso regression. Finally, we evaluate the results using mean squared error (MSE). Our model demonstrates effective sequence data prediction and classification with high stability. Nevertheless, a drawback associated with this is the extended time required to accomplish the tasks. This paper proposes a new ESN model based on memristors, which simulates the USWS characteristics of the brain using the new intelligent component memristor. It addresses the architectural limitations of ESN and promotes the further development of memristor neural networks and brain-inspired models.

Performance Improvement of Memristor-based Echo State Networks by Optimized Programming Scheme

Aging Aware Retraining for Memristor-based Neuromorphic Computing

Memristor-Based Echo State Network with Online Least Mean Square

A Novel Memristors Based Echo State Network Model Inspired by the Brain’s Uni-hemispheric Slow-Wave Sleep Characteristics

The Ouroboros of Memristors: Neural Networks Facilitating Memristor Programming

An Efficient Programming Framework for Memristor-based Neuromorphic Computing

Memristor-based Mixed Training Scheme with Multi-level Sparsification

Dynamic Memristor Based Cyclic Echo State Network for Time Series Prediction

Convolutional Echo‐State Network with Random Memristors for Spatiotemporal Signal Classification

Performance Optimization Of Echo State Networks Through Principal Neuron Reinforcement

Reduction 93.7% time and power consumption using a memristor-based imprecise gradient update algorithm

Memristor-based Echo State Network and Prediction for Time Series

A Novel Memristors Based Echo State Network Model Inspired by the Brain's Uni-hemispheric Slow-Wave Sleep Characteristics

SSM: a High-Performance Scheme for in Situ Training of Imprecise Memristor Neural Networks

Optimizing Supervised Learning of Deep Spiking Neural Network towards Memristor Crossbar Implementation.

A Physics-Based Predictive Model for Pulse Design to Realize High-Performance Memristive Neural Networks

A Tuning-Free Hardware Reservoir Based on MOSFET Crossbar Array for Practical Echo State Network Implementation

A Variation Tolerant Scheme for Memristor Crossbar Based Neural Network Designs Via Two-Phase Weight Mapping and Memristor Programming

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

Reliability-Driven Neural Network Training for Memristive Crossbar-Based Neuromorphic Computing Systems

Mitigate IR-Drop Effect by Modulating Neuron Activation Functions for Implementing Neural Networks on Memristor Crossbar Arrays