Bipolar Resistive Switching in TiO₂ Artificial Synapse Mimicking Pavlov’s Associative Learning

Abstract

Memristive devices are among the most emerging electronic elements to realize artificial synapses for neuromorphic computing (NC) applications and have potential to replace the traditional von-Neumann computing architecture in recent times. In this work, pulsed laser deposition-manufactured Ag/TiO2/Pt memristor devices exhibiting digital and analog switching behavior are considered for NC. The TiO2 memristor shows excellent performance of digital resistive switching with a memory window of order ∼103. Furthermore, the analog resistive switching offers multiple conductance levels supporting the development of the bioinspired synapse. A possible mechanism for digital and analog switching behavior in our device is proposed. Remarkably, essential synaptic functions such as pair-pulse facilitation, long-term potentiation (LTP), and long-term depression (LTD) are successfully realized based on the change in conductance through analog memory characteristics. Based on the LTP-LTD, a neural network simulation for the pattern recognition task using the MNIST data set is investigated, which shows a high recognition accuracy of 95.98%. Furthermore, more complex synaptic behavior such as spike-time-dependent plasticity and Pavlovian classical conditioning is successfully emulated for associative learning of the biological brain. This work enriches the TiO2-based resistive random-access memory, which provides information about the simultaneous existence of digital and analog behavior, thereby facilitating the further implementation of memristors in low-power NC.

Thermally Driven Multilevel Non-Volatile Memory with Monolayer MoS2 for Brain-Inspired Artificial Learning

“…The variable ω and the maximum ( G Max ) and minimum ( G Min ) conductance determine the non-linearity coefficients (α) for potentiation (α p ) and depression (α d ), which are 0.64 and 0.41, respectively. These parameters play a crucial role in the off-chip training procedure for pattern recognition in neural network training …”

Reconfigurable Low-Power TiO2 Memristor for Integration of Artificial Synapse and Nociceptor

“…To quantify the synaptic weight change, the ratio of the final current ( I f ) to the initial current ( I i ) (defined in Figure a) is calculated for all the values of Δ t . For Δ t = 1 μs (1 MHz), the ratio is maximum, while it is minimum for Δ t = 1 ms; the experimental data can be well fitted with the double exponential decay function of the form where C 1 and C 2 are the facilitation factors and τ 1 and τ 2 are the fast and slow decay time constants, respectively. The fitting parameters τ 1 and τ 2 are found to be 0.03 and 0.214 ms, respectively.…”

“…Over the past few years, a surge in research interest is evidenced in neuromorphic computing architecture based on the human brain encompassing all the basic requisites of AI/ML such as cognitive tasks, fault tolerance with superior processing speed, high precision, and very low power consumption. − Synapse, the junction between two neurons, is the basic element in the brain responsible for memory, processing, and learning ability . Artificial synapses based on two terminal memristors can precisely emulate the complex dynamics of Ca + , Na + ions in the biological synapse involved in synaptic plasticity. , In particular, the multi-level switching capability of the memristors endorses vector matrix multiplication in the hardware implementation of artificial neural networks (ANN). − The short-term and long-term retention loss in memristors can be used to emulate synaptic functions such as pair pulse facilitation (PPF), potentiation, depression, spike time-dependent plasticity (STDP), Pavlov’s associative learning, which has direct application in image processing, sparse coding, data classification, and so on. − The stochasticity in the switching behavior can be used as a true random number generator for data encryption and cybersecurity applications. , All these devices are based on drift memristors that mainly exhibit non-volatile switching (NVS) behavior, which cannot faithfully emulate the short-term plasticity of the synapse.…”

Memristor-based neural networks: a bridge from device to artificial intelligence

“…Recently, a study has revealed the signal simulation behavior of a TiO 2 -based memristor in neuromorphic computing. 104 In this work, the Ag/TiO 2 /Pt memristor was fabricated using pulsed laser deposition, resulting in excellent digital and analog switching behavior with a memory window of about ∼10 3 . This memristor could provide multiple conductance levels and successfully realize the basic synaptic function.…”