Research progress of artificial neural systems based on memristors

In the biological nervous system, the integration and cooperation of parallel system of receptors, neurons, and synapses allow efficient detection and processing of intricate and disordered external information. Such systems acquire and process environmental data in real‐time, efficiently handling complex tasks with minimal energy consumption. Memristors can mimic typical biological receptors, neurons, and synapses by implementing key features of neuronal signal‐processing functions such as selective adaption in receptors, leaky integrate‐and‐fire in neurons, and synaptic plasticity in synapses. External stimuli are sensitively detected and filtered by “artificial receptors,” encoded into spike signals via “artificial neurons,” and integrated and stored through “artificial synapses.” The high operational speed, low power consumption, and superior scalability of memristive devices make their integration with high‐performance sensors a promising approach for creating integrated artificial sensory systems. These integrated systems can extract useful data from a large volume of raw data, facilitating real‐time detection and processing of environmental information. This review explores the recent advances in memristor‐based artificial sensory systems. The authors begin with the requirements of artificial sensory elements and then present an in‐depth review of such elements demonstrated by memristive devices. Finally, the major challenges and opportunities in the development of memristor‐based artificial sensory systems are discussed.

The ongoing global energy crisis has heightened the demand for low-power electronic devices, driving interest in neuromorphic computing inspired by the parallel processing of human brains and energy efficiency. Reconfigurable memristors, which integrate both volatile and non-volatile behaviors within a single unit, offer a powerful solution for in-memory computing, addressing the von Neumann bottleneck that limits conventional computing architectures. These versatile devices combine the high density, low power consumption, and adaptability of memristors, positioning them as superior alternatives to traditional complementary metal-oxide-semiconductor (CMOS) technology for emulating brain-like functions. Despite their potential, studies on reconfigurable memristors remain sparse and are often limited to specific materials such as Mott insulators without fully addressing their unique reconfigurability. This review specifically focuses on reconfigurable memristors, examining their dual-mode operation, diverse physical mechanisms, structural designs, material properties, switching behaviors, and neuromorphic applications. It highlights the recent advancements in low-power-consumption solutions within memristor-based neural networks and critically evaluates the challenges in deploying reconfigurable memristors as standalone devices or within artificial neural systems. The review provides in-depth technical insights and quantitative benchmarks to guide the future development and implementation of reconfigurable memristors in low-power neuromorphic computing.

Editorial

The visceral functions of the nervous system are so intimately related to the somatic and higher functions of the nervous system that the selection of papers for an annual review is quite difficult. For example, acetylcholine acts at central, somatic, and visceral neuroeffector junctions and on struc­ tures having no nerves. Furthermore, acetylcholine affects the heart and pe­ ripheral circulation and is involved in the functioning of visceral receptors. Where, then, shall the studies on the action, formation, and destruction of this ubiquitous agent be considered? Some are reviewed here. A new drug is introduced to control hypertension but soon ends up being used primarily in psychiatry. Does this drug affect the visceral functions of the nervous sys­ tem? Should papers concerning its central effects be reviewed here? The an­ swer in this case has been, in general, no. The follo wing, therefore, is a short review of some of the literature on some of the visceral functions of the nervous system. Posterior pit uitary.- Sir Henry Dale (1), in the Banting Memorial Lec­ ture, 1954, informally reviewed six decades of study of the posterior lobe of the pituitary. In describing the neural control of hormone production Dale draws an analogy between the hypothalamicohyp ophyseal system and the better defined neurohumoral transmitter functions of the motor nervous system. He cautions, however, against regarding this analogy as completely true. Ethanol, in the form of 120 ml. whiskey administered orally, increased urine output and free water clearance, and decreased the sodium, potassium, and chloride excretion in normal males [Rubini, Kleeman & Lamdin (2)]. The authors believe these results substantiate the idea that ethanol inhibits the supraopticohypop hyseal

Novel nano-scale communication techniques are inspired by biological systems. Neuro-spike communication is an example of this communication paradigm which transfers vital information about external and internal conditions of the body through the nervous system. The analysis of this communication paradigm is beneficial to exploit in the artificial neural systems where nano-machines are linked to neurons to treat the neurodegenerative diseases. In these networks, nano-machines are used to replace the damaged segments of the nervous system and they exactly behave like biological entities. In neuro-spike communication, neurons / nano-machines exploit the electro-chemical spikes and molecular communication to transfer information. This communication paradigm can be divided into three main parts, namely the axonal pathway, the synaptic transmission, and the spike generation. In this paper, we focus on the axonal transmission part as a separate channel since the capacity of the axonal pathway has a significant effect on the capacity of neuro-spike communication channel. In thinner axons, the capacity of this part is the bottleneck of the neuro-spike communication channel capacity. Hence, we investigate the restricting factors of the axonal transmission which limit its capacity. We derive the capacity of single-input single-output and multiple-input single-output (MISO) axonal channels. In the MISO case, we investigate the effect of the correlation among inputs on the channel capacity. Moreover, we derive a closed form description for the optimum value of the input spike rate to maximize the capacity of the axonal channel when the information is encoded by firing rate of neurons / nano-machines.

The human skin is a unique organ that embeds multiple functions that no artificial systems can currently replicate. Advances in materials science and engineering are driving the design of electronic skins—large-area sensor arrays that mimic some sensory modalities and have the soft, elastic form of natural skin. Here, we focus on electronic skins designed to be worn on the human body for healthcare monitoring or prosthetic applications. The primary sensing modalities are mechanical, thermal, and electrophysiological. We review key materials and associated designs needed to manufacture electronic devices that can conform to the human body and move along with it. Electronic skins offer exciting opportunities for human–machine interfaces.

EDITORIAL article Front. Neurorobot., 13 September 2013 https://doi.org/10.3389/fnbot.2013.00013

The evolution of communication networks and information systems, to support wireless access, cloud and grid computing, and big data, provides great business opportunities. However, it also generates a new trend of sophisticated network threats and offers several challenges in securing information and systems confidentiality, integrity and availability. The traditional techniques used by security experts are mostly static and lack the much needed characteristics of adaptation and self-organization, computational efficiency and error resilience to deal with evolving attacks. The inherent characteristics of computational intelligence (CI) paradigms provide a promising alternative that has gained popularity resulting in significant applications in information security. There is a plethora of CI paradigms commonly used in this domain including artificial neural networks, evolutionary computing, fuzzy systems, and swarm intelligence. This chapter provides an overview of the widely-recognized CI paradigms and shades the light on some of their potential applications in information security.

The visual perception system is the most important system for human learning since it receives over 80% of the learning information from the outside world. With the exponential growth of artificial intelligence technology, there is a pressing need for high-energy and area-efficiency visual perception systems capable of processing efficiently the received natural information. Currently, memristors with their elaborate dynamics, excellent scalability, and information (e.g., visual, pressure, sound, etc.) perception ability exhibit tremendous potential for the application of visual perception. Here, we propose a fully memristor-based artificial visual perception nervous system (AVPNS) which consists of a quantum-dot-based photoelectric memristor and a nanosheet-based threshold-switching (TS) memristor. We use a photoelectric and a TS memristor to implement the synapse and leaky integrate-and-fire (LIF) neuron functions, respectively. With the proposed AVPNS we successfully demonstrate the biological image perception, integration and fire, as well as the biosensitization process. Furthermore, the self-regulation process of a speed meeting control system in driverless automobiles can be accurately and conceptually emulated by this system. Our work shows that the functions of the biological visual nervous system may be systematically emulated by a memristor-based hardware system, thus expanding the spectrum of memristor applications in artificial intelligence.

A neural network classification method has been developed as an alternative approach to the search/organization problem of large modular databases. Two artificial neural systems have been implemented on a Cray for rapid protein/nucleic acid classification of unknown sequences. The system employs a n-gram hashing function for sequence encoding and modular backpropagation networks for classification. The protein system has achieved a 82 to 100% sensitivity at a speed that is about an order of magnitude faster than other search methods. With the rapid accumulation of sequences, the saving in time will become increasingly significant. The pilot nucleic acid system showed a 96% classification accuracy. The software tool would be valuable for the organization of molecular sequence databases and is generally applicable to any databases that are organized according to family relationships.

The article discusses the main trends in the development of artificial intelligence systems and robotics (AI&R). The main question that is considered in this context is whether artificial systems are going to become more and more anthropomorphic, both intellectually and physically. In the article, the author analyzes the current state and prospects of technological development of artificial intelligence and robotics and also determines the main aspects of the impact of these technologies on society and economy, indicating the geopolitical strategic nature of this influence. The author considers various approaches to the definition of artificial intelligence and robotics, focusing on the subject-oriented and functional ones. In the article, AI&R abilities and human abilities are compared in such areas as categorization, pattern recognition, planning and decision making, etc. Based on this comparison, it is concluded in which areas AI&R’s performance is inferior to a human and in which cases it is superior to one. The modern achievements in the field of robotics and artificial intelligence create the necessary basis for further discussion of the applicability of goal setting in engineering, in the form of a Turing test. It is shown that development of AI&R is associated with certain contradictions that impede the application of Turing’s methodology in its usual format. The basic contradictions in the development of AI&R technologies imply that there is to be a transition to a post-Turing methodology for assessing engineering implementations of AI&R. In such implementations, on the one hand, the “Turing wall” is removed, and on the other hand, artificial intelligence gets its physical implementation.

Spiking neural networks (SNNs) are the basis for many energy-efficient neuromorphic hardware systems. While there has been substantial progress in SNN research, artificial SNNs still lack many capabilities of their biological counterparts. In biological neural systems, memory is a key component that enables the retention of information over a huge range of temporal scales, ranging from hundreds of milliseconds up to years. While Hebbian plasticity is believed to play a pivotal role in biological memory, it has so far been analyzed mostly in the context of pattern completion and unsupervised learning in artificial and SNNs. Here, we propose that Hebbian plasticity is fundamental for computations in biological and artificial spiking neural systems. We introduce a novel memory-augmented SNN architecture that is enriched by Hebbian synaptic plasticity. We show that Hebbian enrichment renders SNNs surprisingly versatile in terms of their computational as well as learning capabilities. It improves their abilities for out-of-distribution generalization, one-shot learning, cross-modal generative association, language processing, and reward-based learning. This suggests that powerful cognitive neuromorphic systems can be built based on this principle.

Neuromorphic skin is an emerging electronic skin that demonstrates sensory, memory, learning, and motor responses in a neuromorphic way by using advanced artificial synaptic devices that emulate the biological nervous system and its neural plasticity. Many types of artificial synapses that emulate brain‐inspired computing have been developed and are being integrated in electronic skin to demonstrate artificial sensory and motor nervous systems. The neuromorphic skin, which operates using voltage spikes thereof, precisely mimics the energy‐efficient information processing of biological nervous system in the brain and body and enables the implementation of power‐ and time‐efficient cognitive sensors and computing, biomimetic robotics and electronics, and biocompatible neuroprosthetics. Here, biological synapses and emerging artificial synapses are introduced, and recent research is reviewed on neuromorphic skins that use artificial synapses for sensing, memory, learning, and motor systems. Finally, the outlook and challenges to be addressed are provided for future advanced neuromorphic skin in which all neuromorphic parts are integrated with brain‐inspired computing and bioinspired robotics.

Artificial Systems (ANS) technology has shown great promise as a means of addressing some of the very complex pattern matching problems that have heretofore eluded solution by computer; examples of such problems include resource allocation for battle management, real-time target identification and classification, and other open-world problems. However, it is difficult to implement a realistic ANS solution to these problems due to the inherent differences between ANS architectures and the computers used to simulate them. Eventually, this problem will be solved by the development of integrated circuits designed specifically for ANS applications. In the interim, researchers and applications developers are faced with having to use software techniques on uniprocessor systems to simulate the parallelism of an ANS model. This approach allows development of proof-of-concept applications for the technology, but generally precludes the integration of ANS processing techniques into operational systems. To alleviate this situation, a parallel processing computer system with an architecture specifically tailored to ANS processing requirements has been designed and implemented at Ford Aerospace. This machine, dubbed the Neural Emulation Tool (NET), is based on an extensible array processor design that allows the user to add computer and/or memory capacity to the system in a modular fashion. The parallelism provided by this architecture, coupled with the ANS simulation software developed for NET, provides the computer resources needed to implement large scale ANS simulations in laboratory or operational applications. This paper describes the problems associated with simulating an ANS on traditional computing platforms, including many parallel processing systems, and shows how the architecture of the NET computer addresses these issues to provide fast and flexible ANS simulations in a small computer system.

Chapter 5 - Limitations of Small Devices and Large Systems