Unsupervised Learning Implemented by Ti₃C₂-MXene-Based Memristive Neuromorphic System

Abstract

The neuromorphic hardware system has been a promising candidate for future computing architectures, as it enables adaptive learning at low energy and area consumption. However, hardware implementation of unsupervised learning is still not well-studied. In this work, we design a memristor-based hardware system to realize mean-shift (an unsupervised learning algorithm). A crossbar array of Ti3C2-MXene-based memristors is used to perform a multiply accumulation operation and conductance training. In simulations with device properties, mean-shift-algorithm-based target tracking is successfully demonstrated with comparable accuracy to the software-based result. This work provides an approach to realize unsupervised learning with a memristive neuromorphic system.

Preparation of MXene-based hybrids and their application in neuromorphic devices

“…Interestingly, the MXene-based device exhibited lower operation voltage, higher stability, more resistance states and stronger endurance. In a separate study involving unsupervised learning networks, MXene/SiO 2 configuration-based memristor devices could be used to construct hardware neuromorphic systems [123].…”

Memristive devices based hardware for unlabeled data processing

“…They have a unique characteristic that their internal resistance (or conductance) state is determined by the history of applied voltages and currents. These nanoscale devices feature good scalability, stacking-ability, and other intriguing properties that exceed conventional integrated circuit technology [49][50][51][52][53][54]. Besides, there exists a high similarity between the variable synaptic strengths of biological synapses and tunable internal resistance states of memristive devices.…”

Artificial Neurons Based on Ag/V2C/W Threshold Switching Memristors

“…Among them, MXene, a family of two-dimensional (2D) transition-metal carbides, carbonitrides, and nitrides, has shown interesting semiconductor characteristics owing to their abundant electrochemically active surfaces [29][30][31][32]. For example, previous reports have been investigated on the MXene (Ti 3 C 2 )-based memristors to realize fast pulse modulation time and emulation of neuromorphic behaviors [20,[33][34][35][36]. In particular, it is reported that three-atom-type MXene (e.g., V 2 C) exhibited ultra-low power, more stable endurance, and multiple synaptic functions, i.e., short-term plasticity (STP), long-term plasticity (LTP), spike-timing-dependent plasticity (STDP), and spike-rate-dependent plasticity (SRDP), due to its more stable atomic structure and higher conductivity [37][38][39].…”