Neuromorphic Engineering for Hardware Computational Acceleration and Biomimetic Perception Motion Integration

The nonvolatile and continuous tuning of device conductance is the crucial requirement when developing novel materials and devices as artificial synapses. In this work we demonstrate that by exploiting various kinds of mechanisms such as ferroelectric tuning, floating gate controlling, ferroelectric-like transition in van der Waals layered 2-dimensional materials, electrokinetic phenomenon in nanofluid, etc. low dimensional devices have been successfully fabricated as memristive synapses. The advantages and challenges of using these low dimensional materials and devices as electronic synapses are illustrated with the hope of providing guidelines for the future development of this promising topic.

With the advent of Big Data, traditional digital computing is struggling to cope with intricate tasks related to data classification or pattern recognition. To mitigate this limitation, software-based neural networks are implemented, but they are run in conventional computers whose operation principle (with separate memory and data-processing units) is highly inefficient compared to the human brain. Brain-inspired in-memory computing is achieved through a wide variety of methods, for example, artificial synapses, spiking neural networks, or reservoir computing. However, most of these methods use materials (e.g., memristor arrays, spintronics, phase change memories) operated with electric currents, resulting in significant Joule heating effect. Tuning magnetic properties by voltage-driven ion motion (i.e., magnetoionics) has recently emerged as an alternative energy-efficient approach to emulate functionalities of biological synapses: potentiation/depression, multilevel storage, or transitions from short-term to long-term plasticity. In this perspective, the use of magnetoionics in neuromorphic applications is critically reviewed, with emphasis on modulating synaptic weight through: 1) control of magnetization by voltage-induced ion retrieval/insertion; and 2) control of magnetic stripe domains and skyrmions in gated magnetic thin films adjacent to solid-state ionic supercapacitors. The potential prospects in this emerging research area together with a forward-looking discussion on future opportunities are provided.

Mimicking Pavlovian conditioning by memristive synapse is significant to implement neuromorphic computing at the hardware level. In this work, we demonstrated the Pavlovian conditioning based on the artificial synapse architecture of one-transistor/one-resistor (1T1R), which included an AgInSbTe/α-C-based memristor as a variable resistance and an N-MOS transistor. Thanks to stable resistance switching behavior of memristor and outstanding controllability of device conductance by transistor gating of 1T1R, the experimental demonstration of the acquisition and extinction of Pavlovian conditioning were realized. Moreover, the temporal relation between the conditioned and unconditioned stimuli was also established in which the memory time of associative learning decreased with the increase in the interval of two stimuli. This work provided an idea to biorealistically mimic the Pavlovian conditioning, paving the way for memristive neuromorphic computing.

All-in-one metal-oxide heterojunction artificial synapses for visual sensory and neuromorphic computing systems

The state-of-the-art artificial intelligence technologies mainly rely on deep learning algorithms based on conventional computers with classical von Neumann computing architectures, where the memory and processing units are separated resulting in an enormous amount of energy and time consumed in the data transfer process. Inspired by the human brain acting like an ultra-highly efficient biological computer, neuromorphic computing is proposed as a technology for hardware implementation of artificial intelligence. Artificial synapses are the main component of a neuromorphic computing architecture. Memristors are considered to be a relatively ideal candidate for artificial synapse applications due to their high scalability and low power consumption. Oxides are most widely used in memristors due to the ease of fabrication and high compatibility with complementary metal-oxide-semiconductor processes. However, oxide memristors suffer from unsatisfactory stability and reliability. Oxide-based hybrid structures can effectively improve the device stability and reliability, therefore providing a promising prospect for the application of oxide memristors to neuromorphic computing. This work reviews the recent advances in the development of hybrid oxide memristive synapses. The discussion is organized according to the blending schemes as well as the working mechanisms of hybrid oxide memristors.

Spintronic devices for high-density memory and neuromorphic computing – A review

Neuromorphic computing has seen great interest as leaps in artificial intelligence (AI) applications have exposed limitations due to heavy memory access, with the von Neumann computing architecture. The parallel in‐memory computing provided by neuromorphic computing has the potential to significantly improve latency and power consumption. Key to analog neuromorphic computing hardware are memristors, providing non‐volatile multistate conductance levels, high switching speed, and energy efficiency. Ferroelectric tunnel junction (FTJ) memristors are prime candidates for this purpose, but the impact of the particular characteristics for their performance upon integration into large crossbar arrays, the core compute element for both inference and training in deep neural networks, requires close investigation. In this work, a W/Hf x Zr1−x O2/TiN FTJ with 60 programmable conductance states, a dynamic range (DR) up to 10, current density >3 A m−2 at V read = 0.3 V and highly nonlinear current–voltage (I–V) characteristics (>1100) is experimentally demonstrated. Using a circuit macro‐model, the system level performance of a true crossbar array is evaluated and a 92% classification accuracy of the modified nation institute of science and technology (MNIST) dataset is achieved. Finally, the low on conductance in combination with the highly nonlinear I–V characteristics enable the realization of large selector‐free crossbar arrays for neuromorphic hardware accelerators.

Due to the voltage driven switching at low voltages combined with nonvolatility of the achieved polarization state, ferroelectric materials have a unique potential for low power nonvolatile electronic devices. The competitivity of such devices is hindered by compatibility issues of well-known ferroelectrics with established semiconductor technology. The discovery of ferroelectricity in hafnium oxide changed this situation. The natural application of nonvolatile devices is as a memory cell. Nonvolatile memory devices also built the basis for other applications like in-memory or neuromorphic computing. Three different basic ferroelectric devices can be constructed: ferroelectric capacitors, ferroelectric field effect transistors and ferroelectric tunneling junctions. In this article first the material science of the ferroelectricity in hafnium oxide will be summarized with a special focus on tailoring the switching characteristics towards different applications.The current status of nonvolatile ferroelectric memories then lays the ground for looking into applications like in-memory computing. Finally, a special focus will be given to showcase how the basic building blocks of spiking neural networks, the neuron and the synapse, can be realized and how they can be combined to realize neuromorphic computing systems. A summary, comparison with other technologies like resistive switching devices and an outlook completes the paper.

Ferrimagnetic materials with antiparallel exchange coupling, exhibit spin‐orbit‐torque‐induced dynamics, offering an emerging platform for realizing neuromorphic devices, such as artificial synapses. However, the state‐of‐the‐art artificial synapses based on ferrimagnet suffer from poor analog switching linearity, which serves as a bottleneck for achieving complex tasks with high accuracy in neuromorphic computing. Here, an artificial synapse is reported with high‐weight update linearity in a compensated ferrimagnetic crossbar device. In particular, the linear weight update of the synapses is enhanced by engineering the current density distribution. Using experimentally derived device parameters, handwritten digit recognition can be achieved with an accuracy of over 95% in a three‐layer fully connected artificial neural network. The work provides a universal method to improve the synaptic linearity, which also paves the way for applying the spin‐orbit device in neuromorphic computing.

The human brain, with high energy‐efficient and parallel processing ability, inspires to mitigate power issues perplexing von Neumann architecture. As one of the essential components constructing the human brain, the emulation of biological synapses exploiting electronic devices consuming power at a biological level lays the foundation for the implementation of energy‐efficient neuromorphic computing. Besides, signal matching between biologically‐related stimuli and the driving voltage of artificial synapses helps to realize intelligent neuromorphic interfaces and sustainable energy. Here, ultra‐sensitive artificial synapse stimulated at 1 mV with energy consumption of 132 attojoule/synaptic event is demonstrated. Biological signal matching and low power application are realized simultaneously based on sodium acetate (NaAc) doped polyvinyl alcohol (PVA) electrolyte. The biphasic current, which comprises the electrical‐ and ion‐mediation current component, contributes to enrich synaptic functions compared to monophasic synaptic behavior. Moreover, freestanding NaAc‐doped PVA membrane functions as both dielectric layer and mechanical support and facilitates to achieve flexible, transferable artificial synapse, which maintains functional stability at an ultralow voltage and power even after bending tests. Thus, encompassing superior sensitivity, low energy, and multiple functionalities with flexible, self‐supported, biocompatible property, takes a step to construct energetically‐efficient, complex neuromorphic systems for wearable, implantable medicines as well as smart bio‐electronic interfaces.

Point-contact enabled reliable and low-voltage memristive switching and artificial synapse from highly transparent all-oxide-integration

Artificial synapse having good linearity is crucial to achieve an efficient learning process in neuromorphic computing. It is found that the synaptic linearity can be enhanced by engineering the doping region across the switching layer. The nonlinearity of potentiation and depression of the pure device is 36% and 91%, respectively; meanwhile, the nonlinearity after doping can be suppressed to be 22% (potentiation) and 60% (depression). Henceforth, the learning accuracy of the doped device is 91% with only 13 iterations; meanwhile, the pure device is 78%. A detailed conduction mechanism to understand this phenomenon is proposed.

In [1] a method to accurately compute the Lyapunov Characteristic Exponents of continuous dynamical systems of arbitrary dimensions was presented. However, it can be computationally expensive, because it requires the computation of the time derivatives of the entries of the exponential of a skew-symmetric matrix. In this paper, we present an implementation of the method in [1] that takes advantage of the fact that some of the computations can be done in parallel. The speedup in the computations depends on the number of CPU cores used and the computer memory. Numerical simulations show improvements in efficiency when using the parallel implementation. Our implementation retains the accuracy of the method in [1] with the added advantage of a speedup in computations. Numerical simulation results are presented for a dynamical system of dimension seven and one of dimension forty-nine.

Von Neumann architecture-based computing systems are facing a von Neumann bottleneck owing to data transfer between separated memory and processor units. In-memory computing (IMC), on the other hand, reduces energy consumption and improves computing performance. This study explains an 8+T SRAM IMC circuit based on 8+T differential SRAM (8+T SRAM) and proposes 8+T SRAM-based IMC full adder (FA) and 8+T SRAM-based IMC approximate adder, which are based on the 8+T SRAM IMC circuit. The 8+T SRAM IMC circuit performs SRAM read and bitwise operations simultaneously and performs each logic operation parallelly. The proposed IMC FA and the proposed IMC approximate adder can be applied to a multi-bit adder. The two adders are based on the 8+T SRAM IMC circuit and thus read and compute simultaneously. In this study, the 8+T SRAM IMC circuit was applied to the adder, leveraging its ability to perform read and logic operations simultaneously. According to the performance in this study, the 8+T SRAM IMC circuit, proposed FA, proposed RCA, and proposed approximated adder are good candidates for IMC, which aims to reduce energy consumption and improve overall performance.

New machine learning algorithms such as deep neural networks and the availability of large datasets have created a large drive towards new types of hardware capable of executing these algorithms with higher energy-efficiency. Recently, silicon photonics has emerged as a promising hardware platform for neuromorphic computing due to its inherent capability to process linear and non-linear operations and transmit a high bandwidth of data in parallel. At Hewlett Packard Labs, an energy-efficient dense-wavelength division multiplexing (DWDM) silicon photonics platform has been developed as the underlying foundation for innovative neuromorphic computing architectures. The latest research on our silicon photonic neuromorphic platform will be presented and discussed.