Y2O3-based Memristive Crossbar Array for Synaptic Learning

Here, we report a fabrication of Y2O3-based memristive crossbar array along with an analytical model to evaluate the performance of such memristive array system to understand the forgetting and retention behavior in the neuromorphic computation. The developed analytical model is able to simulate the highly-dense memristive crossbar array based neural network of biological synapses. These biological synapses control the communication efficiency between neurons and can implement the learning capability of the neurons. During electrical stimulation of the memristive devices, the memory transition is exhibited along with the number of applied voltage pulses which is analogous to the real human brain functionality. Further, to obtain the forgetting and retention behavior of the memristive devices, a modified window function equation is proposed by incorporating two novel internal state variables in the form of forgetting rate and retention. The obtained results confirm that the effect of variation in electrical stimuli on forgetting and retention as similar to the biological brain. Therefore, the developed analytical memristive model further can be utilized in the memristive system to develop real-world applications in neuromorphic domains.

Dynamic resistive switching devices for neuromorphic computing

Neuromorphic systems that can emulate the structure and the operations of biological neural circuits have long been viewed as a promising hardware solution to meet the ever-growing demands of big-data analysis and AI tasks. Recent studies on resistive switching or memristive devices have suggested such devices may form the building blocks of biorealistic neuromorphic systems. In a memristive device, the conductance is determined by a set of internal state variables, allowing the device to exhibit rich dynamics arising from the interplay between different physical processes. Not only can these devices be used for compute-in-memory architectures to tackle the von Neumann bottleneck, the switching dynamics of the devices can also be used to directly process temporal data in a biofaithful fashion. In this Review, we analyze the physical mechanisms that govern the dynamic switching behaviours and highlight how these properties can be utilized to efficiently implement synaptic and neuronal functions. Prototype systems that have been used in machine learning and brain-inspired network implementations will be covered, followed with discussions on the challenges for large scale implementations and opportunities for building bio-inspired, highly complex computing systems.

Sample preparation of dense granular materials Influence of void ratio e and coordination number Z* on the mechanical behaviour at failure

This paper presents a Discrete Element Method (DEM) study of assemblies of 5041 frictional discs, with periodic boundary conditions and confined by an external isotropic load. In order to generate samples with different internal state variables like the void ratio and coordination number, we present two different numerical procedures. The first technique, which has been widely used in the literature for many years, consists in controlling the coefficient of friction between particles to adjust the density of the samples, which directly influences the coordination number. The second technique is inspired by the previous one but adds an extra step of dynamic mixing with intergranular contacts lubrication. This makes it possible to control quasi independently the void ratio and the coordination number in the case of dense samples. These two types of samples are subjected to simple shear to analyse the influence of the sample preparation procedure on their macroscopic mechanical behaviour.