Implementation of Binary Stochastic STDP Learning Using Chalcogenide-Based Memristive Devices

Abstract

The emergence of nano-scale memristive devices encouraged many different research areas to exploit their use in multiple applications. One of the proposed applications was to implement synaptic connections in bio-inspired neuromorphic systems. Large-scale neuromorphic hardware platforms are being developed with increasing number of neurons and synapses, having a critical bottleneck in the online learning capabilities. Spiketiming-dependent plasticity (STDP) is a widely used learning mechanism inspired by biology which updates the synaptic weight as a function of the temporal correlation between pre-and postsynaptic spikes. In this work, we demonstrate experimentally that binary stochastic STDP learning can be obtained from a memristor when the appropriate pulses are applied at both sides of the device.

Neuromorphic Synapses with High Switching Uniformity and Multilevel Memory Storage Enabled through a Hf-Al-O Alloy for Artificial Intelligence

“…One can influence the anion migration of oxygen ions using oxide materials, which have a variety of intriguing qualities such as carrier selectivity, nontoxicity, and abundance. These properties make them superior to organic/inorganic, perovskite, and chalcogenide materials for consideration in brain-inspired emulation. − In oxide-based memristive devices, through employing electronic pulses, resistivity/conductivity of the insulating film (sandwiched between two metal electrodes) could be altered by controlling the current limits and stopping voltage, which helps realize biological synaptic functions. − Hafnium oxide (HfO 2 ) has grown in popularity as a result of its demonstrated compatibility with silicon-based complementary metal-oxide-semiconductor (CMOS) technology, excellent resistive switching (RS) characteristics, electrical stability, high ON/OFF ratio, and better controllability in conductance states. − Bilayer memristors are highly useful for realizing synaptic functions by establishing a thin interface/buffer film sandwich between two electrode and insulating layers …”

Stochastic resonance effect in binary STDP performed by RRAM devices

“…STDP is characterized by the time difference between pre-and post-synaptic impulses [3], that determine the direction and magnitude (weight) of the change in synaptic plasticity [4]. In the binary STDP, only '1' or '0' values are valid for each synaptic weight [5]. Thus, binary STDP implemented with memristors involves switching the memristors resistance between two possible resistive states, i.e., HRS (High Resistive State) and LRS (Low Resistive State) (Figure1).…”