Search

Abstract Developing devices with a wide-temperature range persistent photoconductivity (PPC) and ultra-low power consumption remains a significant challenge for optical synaptic devices used in neuromorphic computing. By harnessing the PPC properties in materials, it can achieve optical storage and neuromorphic computing, surpassing the von Neuman architecture-based systems. However, previous research implemented PPC required additional gate voltages and low temperatures, which need additional energy consumption and PPC cannot be achieved across a wide temperature range. Here, we fabricated a simple heterojunctions using zinc(II)-meso-tetraphenyl porphyrin (ZnTPP) and single-walled carbon nanotubes (SWCNTs). By leveraging the strong binding energy at the heterojunction interface and the unique band structure, the heterojunction achieved PPC over an exceptionally wide temperature range (77 K-400 K). Remarkably, it demonstrated nonvolatile storage for up to 2×104 s, without additional gate voltage. The minimum energy consumption for each synaptic event is as low as 6.5 aJ. Furthermore, we successfully demonstrate the feasibility to manufacture a flexible wafer-scale array utilizing this heterojunction. We applied it to autonomous driving under extreme temperatures and achieved as a high impressive accuracy rate as 94.5%. This tunable and stable wide-temperature PPC capability holds promise for ultra-low-power neuromorphic computing.

Poly (butylene adipate-co-terephthalate), PBAT, is one of the most popular plastics nowadays and is becoming a large burden to the environment when PBAT waste is accumulated. Biodegradation of PBAT is an environmentally friendly method to treat PBAT waste. In the past two decades, a couple of microbial strains or mixed strains have been chemically characterized. From this, many PBAT depolymerases have been isolated and further structurally determined. The review we focused on the PBAT depolymerases that have been structurally characterized, including lipases, esterases, and cutinases. We also discussed which shape of the substrate-binding tunnels is beneficial for recognizing and decomposing polymer chains of PBAT, which would be important guidance in directing the designation of PBAT depolymerases.

Abstract Neural circuits with specific structures and diverse neuronal firing features are the foundation for supporting intelligent tasks in biology and are regarded as the driver for catalyzing next-generation artificial intelligence. Emulating neural circuits in hardware underpins engineering highly efficient neuromorphic chips, however, implementing a firing features-driven functional neural circuit is still an open question. In this work, inspired by avoidance neural circuits of crickets, we construct a spiking feature-driven sensorimotor control neural circuit consisting of three memristive Hodgkin-Huxley neurons. The ascending neurons exhibit mixed tonic spiking and bursting features, which are used for encoding sensing input. Additionally, we innovatively introduce a selective communication scheme in biology to decode mixed firing features using two descending neurons. We proceed to integrate such a neural circuit with a robot for avoidance control and achieve lower latency than conventional platforms. These results provide a foundation for implementing real brain-like systems driven by firing features with memristive neurons and put constructing high-order intelligent machines on the agenda.

Artificial intelligence for graph‐structured data has achieved remarkable success in applications such as recommendation systems, social networks, drug discovery, and circuit annotation. Graph convolutional networks (GCNs) are an effective way to learn representations of various graphs. The increasing size and complexity of graphs call for in‐memory computing (IMC) accelerators for GCN to alleviate massive data transmission between off‐chip memory and processing units. However, GCN implementation with IMC is challenging because of the large memory consumption, irregular memory access, and device nonidealities. Herein, a fully binarized GCN (BGCN) accelerator based on computational resistive random‐access memory (RRAM) through software–hardware codesign is presented. The essential operations including aggregation and combination in GCN are implemented on the RRAM crossbar arrays with cooperation between multiply‐and‐accumulation and content‐addressable memory operations. By leveraging the model quantization and IMC on the RRAM, the BGCN accelerator demonstrates less RRAM usage, high robustness to the device variations, high energy efficiency, and comparable classification accuracy compared to the current state‐of‐the‐art GCN accelerators on both graph classification task using the MUTAG and PTC datasets and node classification task using the Cora and CiteSeer datasets. These results provide a promising approach for edge intelligent systems to efficiently process graph‐structured data.

Abstract Combinatorial optimization (CO) has a broad range of applications in various fields, including operations research, computer science, and artificial intelligence. However, many of these problems are classified as nondeterministic polynomial‐time (NP)‐complete or NP‐hard problems, which are known for their computational complexity and cannot be solved in polynomial time on traditional digital computers. To address this challenge, continuous‐time Ising machine solvers have been developed, utilizing different physical principles to map CO problems to ground state finding. However, most Ising machine prototypes operate at speeds comparable to digital hardware and rely on binarizing node states, resulting in increased system complexity and further limiting operating speed. To tackle these issues, a novel device‐algorithm co‐design method is proposed for fast sub‐optimal solution finding with low hardware complexity. On the device side, a piezoelectric lithium niobate (LiNbO3) microelectromechanical system (MEMS) oscillator network‐based Ising machine without second‐harmonic injection locking (SHIL) is devised to solve Max‐cut and graph coloring problems. The LiNbO3 oscillator operates at speeds greater than 9 GHz, making it one of the fastest oscillatory Ising machines. System‐wise, an innovative grouping method is used that achieves a performance guarantee of 0.878 for Max‐cut and 0.658 for graph coloring problems, which is comparable to Ising machines that utilize binarization.

Potherb mustard (Brassica juncea var. multiceps) is one of the most commonly consumed leafy vegetable mustards, either fresh or in pickled form. It is rich in glucosinolates, whose hydrolyzed products confer potherb mustard’s distinctive flavor and chemopreventive properties. In this study, the composition and content of glucosinolates, as well as the hydrolysis pattern of sinigrin were investigated in potherb mustard leaves of different varieties. Variations in the glucosinolate profile and accumulation were observed among the potherb mustard varieties studied, with sinigrin being the predominant one in all varieties, accounting for 81.55% to 97.27%. Sinigrin tended to be hydrolyzed to isothiocyanate (ITC) rather than epithionitrile (EPN) in potherb mustard, while 3-butenyl nitrile (SIN-NIT) could be hardly detected. Transcriptome analysis revealed a higher expression level of numerous genes involved in aliphatic glucosinolate biosynthesis in X11 compared to X57, corresponding to the higher aliphatic glucosinolate accumulation in X11 (91.07 µmol/g) and lower level in X57 (25.38 µmol/g). ESM1 is known to repress nitrile formation and favor isothiocyanate production during glucosinolate hydrolysis. In this study, all four ESM1s showed a higher expression level in X11 compared to X57, which may determine the hydrolysis pattern of sinigrin in potherb mustard. Altogether, our findings shed light on the glucosinolate metabolic pattern in potherb mustard, which will also facilitate the engineering of metabolic pathways at key checkpoints to enhance bioactive compounds for tailored flavor or pharmaceutical needs.

IntroductionThis study explores the emotional impact of religion-related films through a “cinematherapy” lens. It aims to analyze the emotional patterns in a curated selection of religion-related films compared to a broader sample of acclaimed movies using facial recognition with YOLOv5 object detection. The study aims to uncover the potential therapeutic application of religion-related films.MethodsFacial recognition with YOLOv5 object detection was utilized in this study to analyze the emotional patterns in religion-related films. A curated selection of these films was compared to a broader sample of acclaimed movies to identify any distinct emotional trajectories.ResultsThe analysis of the emotional patterns revealed that religion-related films exhibited a subtler range of emotions compared to the broader film spectrum. This finding suggests that these films potentially create a safe space for contemplation, aligning with the profound themes often explored in religion-related films. Interestingly, the emotional arc observed in the films mirrored the spiritual journeys depicted in them. The films started with a low point of separation, transitioned through challenges, and culminated in a peak representing spiritual transformation.DiscussionThese findings suggest promise for the therapeutic application of religion-related films. The muted emotional expression in these films creates a safe space for self-reflection, enabling viewers to connect with the struggles of the characters and explore their own values when faced with complex religious ideas. This emotional engagement may contribute to therapeutic goals such as introspection and personal growth. The study unveils the unique emotional power of religion-related films and paves the way for further research on their potential as therapeutic tools. It emphasizes the need for continued exploration of the emotional impact of these films and their capacity to aid in therapeutic goals.

Abstract Doped‐metal oxide‐based memristors, with the potential for improved switching performance and capability for multi‐bit information storage, are attractive candidates in the implementation of artificial neural network (ANN) hardware systems. However, performance and process considerations such as switching behavior and complementary‐metal‐oxide‐semiconductor (CMOS) process compatibility remain a challenge. This study shows that amorphous Zr‐doped HfO2 (HZO) memristors fabricated via a co‐sputtering approach improve the switching performance by providing a controllable knob to modulate defects in the switching layer. At the same time, it satisfies the CMOS process compatibility requirements for industry adoption. HZO memristors with optimized stoichiometry exhibit 30% reduced switching voltages and 50% faster switching as compared to control HfO2 memristors. Concurrently, this study shows that high linearity analog states tuning is achievable via a programming scheme that utilizes voltage pulses with increasing amplitudes. This study further shows via simulation evaluation that HZO memristors implemented in a self‐organizing‐map (SOM) network for Fashion MNIST database classification, achieve an accuracy of 92% with short training cycles. The results thus pave a potential pathway for further development of CMOS process compatible HZO memristors for use in future storage and computing applications.

Abstract Time‐delayed reservoir computing with marked strengths of friendly hardware implementation and low training cost is regarded as a promising solution to realize time and energy‐efficient time series information processing and thus receives growing attention. However, achieving a sufficient number of reproducible reservoir states remains a significant challenge, which severely limits its computing performance. Here, an electric‐double‐layer‐coupled oxide‐based electrolyte‐gated transistor with a shared gate and varying channel lengths is developed to construct a deep time‐delayed reservoir computing system. A variety of short‐term synaptic responses related to inherent ion‐electron‐coupled dynamics at the electrolyte/channel interface are demonstrated, reflecting a flexibly regulable channel current. Different stable and tunable relaxation responses corresponding to varying channel lengths are obtained to enrich reservoir states combined with virtual nodes ways. The spoken‐digit classification and Hénon map prediction tasks are implemented with high accuracy (≈92.2%) and ultralow normalized root mean square error (≈0.013), respectively, validating the significant improvement of the computing performance by introducing additional relaxation responses. This work opens a promising pathway in exploiting oxide‐based electrolyte‐gated transistors for realizing temporal information processing hardware systems.

Despite the significant progress made in deep learning on digital computers, their energy consumption and computational speed still fall short of meeting the standards for brain-like computing. To address these limitations, reservoir computing (RC) has been gaining increasing attention across communities of electronic devices, computing systems, and machine learning, notably with its in-memory or in-sensor implementation on the hardware–software co-design. Hardware regarded, in-memory or in-sensor computers leverage emerging electronic and optoelectronic devices for data processing right where the data are stored or sensed. This technology dramatically reduces the energy consumption from frequent data transfers between sensing, storage, and computational units. Software regarded, RC enables real-time edge learning thanks to its brain-inspired dynamic system with massive training complexity reduction. From this perspective, we survey recent advancements in in-memory/in-sensor RC, including algorithm designs, material and device development, and downstream applications in classification and regression problems, and discuss challenges and opportunities ahead in this emerging field.

Abstract Neuromorphic computing aims to emulate the computing processes of the brain by replicating the functions of biological neural networks using electronic counterparts. One promising approach is dendritic computing, which takes inspiration from the multi-dendritic branch structure of neurons to enhance the processing capability of artificial neural networks. While there has been a recent surge of interest in implementing dendritic computing using emerging devices, achieving artificial dendrites with throughputs and energy efficiency comparable to those of the human brain has proven challenging. In this study, we report on the development of a compact and low-power neurotransistor based on a vertical dual-gate electrolyte-gated transistor (EGT) with short-term memory characteristics, a 30 nm channel length, a record-low read power of ~3.16 fW and a biology-comparable read energy of ~30 fJ. Leveraging this neurotransistor, we demonstrate dendrite integration as well as digital and analog dendritic computing for coincidence detection. We also showcase the potential of neurotransistors in realizing advanced brain-like functions by developing a hardware neural network and demonstrating bio-inspired sound localization. Our results suggest that the neurotransistor-based approach may pave the way for next-generation neuromorphic computing with energy efficiency on par with those of the brain.

Abstract The development of high-performance oxide-based transistors is critical to enable very large-scale integration (VLSI) of monolithic 3-D integrated circuit (IC) in complementary metal oxide semiconductor (CMOS) backend-of-line (BEOL). Atomic layer deposition (ALD) deposited ZnO is an attractive candidate due to its excellent electrical properties, low processing temperature below copper interconnect thermal budget, and conformal sidewall deposition for novel 3D architecture. An optimized ALD deposited ZnO thin-film transistor achieving a record field-effect and intrinsic mobility (µ FE /µ o ) of 85/140 cm2/V·s is presented here. The ZnO TFT was integrated with HfO2 RRAM in a 1 kbit (32 × 32) 1T1R array, demonstrating functionalities in RRAM switching. In order to co-design for future technology requiring high performance BEOL circuitries implementation, a spice-compatible model of the ZnO TFTs was developed. We then present designs of various ZnO TFT-based inverters, and 5-stage ring oscillators through simulations and experiments with working frequency exceeding 10’s of MHz.

Abstract Flow‐dependent background error covariances estimated from short‐term ensemble forecasts suffer from sampling errors due to limited ensemble sizes. Covariance localization is often used to mitigate the sampling errors, especially for high dimensional geophysical applications. Most applied localization methods, empirical or adaptive ones, multiply the Kalman gain or background error covariances by a distance‐dependent parameter, which is a simple linear filtering model. Here two localization methods based on convolutional neural networks (CNNs) learning from paired data sets are proposed. The CNN‐based localization function (CLF) aims to minimize the sampling error of the estimated Kalman gain, and the CNN‐based empirical localization function (CELF) aims to minimize the posterior error of state variables. These two CNN‐based localization methods can provide localization functions that are nonlinear, spatially and temporally adaptive, and non‐symmetric with respect to displacement, without requiring any prior assumptions for the localization functions. Results using the Lorenz05 model show that CLF and CELF can better capture the structures of the Kalman gain than the best Gaspari and Cohn (GC) localization function and the adaptive reference localization method. For both perfect‐ and imperfect‐model experiments, CLF produces smaller errors of the Kalman gain, prior and posterior than the best GC and reference localization, especially for spatially averaged observations. Without model error, CELF has smaller prior and posterior errors than the best GC and reference localization for spatially averaged observations, while with model error, CELF has smaller prior and posterior errors than the best GC and reference localization for single‐point observations.

Breast cancer (BC) ranks as the highest incidence among cancer types in women all over the world. MicroRNAs (miRNAs) are a class of short endogenous non-coding RNA in cells mostly functioning to silence the target mRNAs. In the current study, a miRNA screening analysis identified miR-186-5p to be downregulated in human breast cancer tumors. Functional studies in vitro demonstrated that overexpression of miR-186-5p inhibited cellular proliferation and induced cell apoptosis in multiple breast cancer cell lines including MDA-MB-231, MCF-7, and BT549 cells. Transplantation of the miR-186-5p-overexpressing MDA-MB-231 cells into nude mice significantly inhibited mammary tumor growth in vivo. Sequence blast analysis predicted annexin A9 (ANXA9) as a target gene of miR-186-5p, which was validated by luciferase reporter assay, QRT-PCR analysis, and western blot. Additional gene expression analysis of clinical tumor samples indicated a negative correlation between miR-186-5p and ANXA9 in human breast cancer. Knockdown of ANXA9 mimicked the phenotype of miR-186-5p overexpression. Reintroduction of ANXA9 back rescued the miR-186-5p-induced cell apoptosis. In addition, miR-186-5p decreased the expression of Bcl-2 and increased the expression of p53, suggesting a mechanism regulating miR-186-5p-induced cellular apoptosis. In summary, our study is the first to demonstrate miR-186-5p-ANXA9 signaling in suppressing human breast cancer. It provided a potential therapeutic target in breast cancer.

Objective To investigate the effects of high-dose hyperbaric trioxygen autologous blood therapy (HOT) on oxygenation index (PaO2/FiO2) and serum inflammatory factors in dogs with acute respiratory distress syndrome (ARDS). Methods Twelve healthy adult beagles were randomly divided into 3 groups (n=4). The blank group was injected with normal saline intravenously. The ARDS model was established by intravenous injection of oleic acid (0.12 mL/kg) in the ARDS group and ARDS+ HOT group. The mark of a successful model is that the oxygen and index (PaO2/FiO2)

Designing efficient photonic neuromorphic systems remains a challenge. Here, the authors develop an in-sensor Reservoir Computing system for multi-tasked pattern classification based on a light-responsive semiconducting polymer (p-NDI) with efficient exciton dissociations, charge trapping capability, and through-space charge-transport characteristics.

Abstract Intrinsic plasticity of neurons, such as spontaneous threshold lowering (STL) to modulate neuronal excitability, is key to spatial attention of biological neural systems. In‐memory computing with emerging memristors is expected to solve the memory bottleneck of the von Neumann architecture commonly used in conventional digital computers and is deemed a promising solution to this bioinspired computing paradigm. Nonetheless, conventional memristors are incapable of implementing the STL plasticity of neurons due to their first‐order dynamics. Here, a second‐order memristor is experimentally demonstrated using yttria‐stabilized zirconia with Ag doping (YSZ:Ag) that exhibits STL functionality. The physical origin of the second‐order dynamics, i.e., the size evolution of Ag nanoclusters, is uncovered through transmission electron microscopy (TEM), which is leveraged to model the STL neuron. STL‐based spatial attention in a spiking convolutional neural network (SCNN) is demonstrated, improving the accuracy of a multiobject detection task from 70% (20%) to 90% (80%) for the object within (outside) the area receiving attention. This second‐order memristor with intrinsic STL dynamics paves the way for future machine intelligence, enabling high‐efficiency, compact footprint, and hardware‐encoded plasticity.

BackgroundCrimes of passion, characterized as unpremeditated impulsive aggression, have garnered increasing attention in recent years. Impulsivity, a major factor in crimes of passion, is also a common feature of various health conditions, including Borderline Personality Disorder (BPD). Childhood maltreatment is considered a significant precursor to BPD and is closely related to impulsivity. Although prior research has affirmed the relationship between impulsivity, childhood maltreatment, BPD, and criminal behavior, few studies have examined these variables’ interconnections within the context of crimes of passion. This study seeks to explore the relationship between childhood maltreatment, BPD, and impulsivity in crimes of passion, assessing the former’s impact on the latter.MethodOur research examined 133 crimes of passion offenders and 149 other offenders from the Shenzhen male prison in China. All 282 participants completed The Abbreviated Version of The Barratt Impulsiveness Scale (ABIS), The UPPS Impulsivity Scale (Urgency, Premeditation, Perseverance, Sensation Seeking), The McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD), and The Childhood Trauma Questionnaire (CTQ).ResultsOur findings indicated that (1) crimes of passion offenders scored significantly higher in emotional neglect, physical neglect, and overall childhood maltreatment than did other offenders, and childhood maltreatment scores were notably higher in the high BPD trait group. (2) Crimes of passion offenders demonstrated significantly elevated impulsivity in attention and nonplanning scales compared to other offenders. Impulsivity scores across all scales were also significantly higher in the high BPD trait group. (3) Emotional neglect was found to significantly influence the urgency scale in crimes of passion offenders. An interaction effect was noted between physical abuse and high BPD traits, leading to increased impulsivity in crimes of passion offenders.ConclusionThis study underscores the predictive roles of childhood maltreatment and BPD in determining impulsivity within the context of crimes of passion.

Recurrent neural networks (RNNs) have proven to be indispensable for processing sequential and temporal data, with extensive applications in language modeling, text generation, machine translation, and time-series forecasting. Despite their versatility, RNNs are frequently beset by significant training expenses and slow convergence times, which impinge upon their deployment in edge AI applications. Reservoir computing (RC), a specialized RNN variant, is attracting increased attention as a cost-effective alternative for processing temporal and sequential data at the edge. RC’s distinctive advantage stems from its compatibility with emerging memristive hardware, which leverages the energy efficiency and reduced footprint of analog in-memory and in-sensor computing, offering a streamlined and energy-efficient solution. This review offers a comprehensive explanation of RC’s underlying principles, fabrication processes, and surveys recent progress in nano-memristive device based RC systems from the viewpoints of in-memory and in-sensor RC function. It covers a spectrum of memristive device, from established oxide-based memristive device to cutting-edge material science developments, providing readers with a lucid understanding of RC’s hardware implementation and fostering innovative designs for in-sensor RC systems. Lastly, we identify prevailing challenges and suggest viable solutions, paving the way for future advancements in in-sensor RC technology.

In the era of the Internet of Things, photonic neuromorphic computing presents a promising method for real-time, local processing of vast quantities of data. However, the rigidity of materials used in such devices can considerably impact performance and longevity when subjected to mechanical deformation. In this study, we introduce a liquid optical memristor (LOM) based on an organic-inorganic hybrid in a liquid state. This novel approach offers programmable optical properties and significant mechanical flexibility thanks to the robust photochromic and capillary effects. We have developed a LOM with a 24 dB cm ^−1 modulation depth and over 3-bit nonvolatile memory states. By controlling the droplet morphology to mimic a synapse-like shape, the LOM can withstand strains over 400% and endure misalignment and bending. Furthermore, our findings substantiate the application of LOM for photonic neuromorphic computing systems, yielding 100% accuracy in pattern recognition. The easily-integratable LOM paves the way for the creation of flexible and wearable photonic neuromorphic computing systems.

The piezoelectric thin film composed of single-crystal lithium niobate (LiNbO3) exhibits a remarkably high electromechanical coupling coefficient and minimal intrinsic losses, making it an optimal material for fabricating bulk acoustic wave resonators. However, contemporary first-order antisymmetric (A1) Lamb mode resonators based on LiNbO3 thin films face specific challenges, such as inadequate mechanical stability, limited power capacity, and the presence of multiple spurious modes, which restrict their applicability in a broader context. In this paper, we present an innovative design for A1 Lamb mode resonators that incorporates a support-pillar structure. Integration of support pillars enables the dissipation of spurious wave energy to the substrate, effectively mitigating unwanted spurious modes. Additionally, this novel approach involves anchoring the piezoelectric thin film to a supportive framework, consequently enhancing mechanical stability while simultaneously improving the heat dissipation capabilities of the core.

Background Early treatment is crucial to the delay of the progression of type 2 diabetes mellitus with stable angina pectoris (T2DM-SAP) , which has poor prognosis, such as high rates of disability and mortality. As traditional Chinese medicine (TCM) has unique advantages in preventing diseases, developing a model with TCM and western medicine factors associated with major adverse cardiovascular and cerebrovascular events (MACCEs) incorporated may be a reliable tool that could be used to predict the risk of MACCEs in patients with T2DM-SAP. Objective To develop and assess the applicability of a risk prediction model for MACCEs in T2DM-SAP patients using identified risk factors associated with MACCEs in this group. Methods Participants were 674 inpatients with T2DM-SAP who received diagnostic and treatment services from The First Affiliated Hospital of Henan University of CM from 2012 to 2019. Through the hospital information system, electronic medical records and follow-up data of these patients were collected, including demographics, clinical characteristics, laboratory parameters, TCM symptoms and syndrome differentiation, and outcome (prevalence of MACCEs) . Patients were classified into a MACCEs group (n=190) and a non-MACCEs group (n=484) by prevalence of MACCEs. Independent risk factors for MACCEs in T2DM with SAP were identified using univariate and multivariate Logistic regression, and used to develop a nomogram-based predictive model. Then the model was internally validated using the bootstrap approach, and its predictive value was estimated using ROC analysis, C-index, calibration plot, Hosmer-Lemeshow test and decision curve analysis. Results Based on the multivariate Logistic regression analysis, the factors associated with MACCEs in T2DM-SAP patients (P30%, indicating that the model may be clinically beneficial. Conclusion The risk predictive model for MACCEs in T2DM-SAP patients was developed using the associated factors (including age, cerebrovascular disease history, serum creatinine, dark purple tongue, decreased tongue coating, thready pulse, and obstruction of collateral channels caused by wind-phlegm) identified by us, which has been proven to have good discrimination, calibration, and clinical effectiveness, and could be used as a tool for assessing the risk of MACCEs in patients with T2DM-SAP.

Abstract Autonomous one‐shot on‐the‐fly learning copes with the high privacy, small dataset, and in‐stream data at the edge. Implementing such learning on digital hardware suffers from the well‐known von‐Neumann and scaling bottlenecks. The optical neural networks featuring large parallelism, low latency, and high efficiency offer a promising solution. However, ex‐situ training of conventional optical networks, where optical path configuration and deep learning model optimization are separated, incurs hardware, energy and time overheads, and defeats the advantages in edge learning. Here, we introduced a bio‐inspired material‐algorithm co‐design to construct a hydrogel‐based optical Willshaw model (HOWM), manifesting Hebbian‐rule‐based structural plasticity for simultaneous optical path configuration and deep learning model optimization thanks to the underlying opto‐chemical reactions. We first employed the HOWM as an all optical in‐sensor AI processor for one‐shot pattern classification, association and denoising. We then leveraged HOWM to function as a ternary content addressable memory (TCAM) of an optical memory augmented neural network (MANN) for one‐shot learning the Omniglot dataset. The HOWM empowered one‐shot on‐the‐fly edge learning leads to 1000× boost of energy efficiency and 10× boost of speed, which paves the way for the next‐generation autonomous, efficient, and affordable smart edge systems.

Abstract The search for high‐performance resistive random‐access memory (RRAM) devices is essential to pave the way for highly efficient non‐Von Neumann computing architecture. Here, it is reported on an alloying approach using atomic layer deposition for a Zn‐doped HfOx‐based resistive random‐access memory (HfZnO RRAM), with improved performance. As compared with HfOx RRAM, the HfZnO RRAM exhibits reduced switching voltages (>20%) and switching energy (>3×), as well as better uniformity both in voltages and resistance states. Furthermore, the HfZnO RRAM exhibits stable retention exceeding 10 years, as well as write/erase endurance exceeding 105 cycles. In addition, excellent linearity and repeatability of conductance tuning can be achieved using the constant voltage pulse scheme, achieving ≈90% accuracy in a simulated multi‐layer perceptron network for the recognition of modified national institute of standards and technology database handwriting. The HfZnO RRAM is also characterized down to the temperature of 4 K, showing functionality and the elucidation of its carrier conduction mechanism. Hence, a potential pathway for doped‐RRAM to be used in a wide range of temperatures including quantum computing and deep‐space exploration is shown.

Abstract The ability to sense light, heat, and touch is vital for human beings, underpinning the interaction between humans and the environment. To mimic the biological perception system, the sensory system converts external light, heat, and mechanical inputs into electrical signals, then processing and storing the data in digital hardware before providing feedback. However, modern digital sensing‐processing systems based on the von Neumann architecture are facing significant challenges in power consumption and latency due to the unprecedented increase in data size and algorithm complexity. A promising solution is to integrate sensors, memory, and control. Here the “state‐of‐the‐art” fusion systems involved in the sensing of visual, olfactory, tactile, visual signals, and control is reviewed. The challenges in high performance and reliability are also discussed.

The unprecedented development of Internet of Things results in the explosion of spatiotemporal signals generated by smart edge devices, leading to a surge of interest in real‐time learning of such data. This imposes a big challenge to conventional digital hardware because of physically separated memory and processing units and the transistor scaling limit. Memristors are deemed a solution for efficient and portable deep learning. However, their ionic resistive switching incurs large programming stochasticity and energy, compromising their advantages in real‐time learning spatiotemporal signals. To address the aforementioned issues, we propose a novel hardware–software codesign. Hardware‐wise, the stochasticity in memristor programming is leveraged to produce random matrices for efficient in‐memory computing. Software‐wise, random convolutional‐pooling architectures are integrated with echo‐state networks that compute with the hardware random matrices and make real‐time learning affordable. The synergy of the hardware and software not only improves the performance over conventional echo‐state networks, that is, 90.94% and 91.67% (compared to baselines 88.33% and 62.50%), but also retains 187.79× and 93.66× improvement of energy efficiency compared to the digital alternatives on the representative Human Activity Recognition Using Smartphones (HAR) and CRICKET datasets, respectively. These advantages make random convolutional echo‐state network (RCESN) a promising solution for the future smart edge hardware.

Artificial neural networks have acquired remarkable achievements in the field of artificial intelligence. However, it suffers from catastrophic forgetting when dealing with continual learning problems, i.e., the loss of previously learned knowledge upon learning new information. Although several continual learning algorithms have been proposed, it remains a challenge to implement these algorithms efficiently on conventional digital systems due to the physical separation between memory and processing units. Herein, a software–hardware codesigned in‐memory computing paradigm is proposed, where a mixed‐precision continual learning (MPCL) model is deployed on a hybrid analogue–digital hardware system equipped with resistance random access memory chip. Software‐wise, the MPCL effectively alleviates catastrophic forgetting and circumvents the requirement for high‐precision weights. Hardware‐wise, the hybrid analogue–digital system takes advantage of the colocation of memory and processing units, greatly improving energy efficiency. By combining the MPCL with an in situ fine‐tuning method, high classification accuracies of 94.9% and 95.3% (software baseline 97.0% and 97.7%) on the 5‐split‐MNIST and 5‐split‐FashionMNIST are achieved, respectively. The proposed system reduces ≈200 times energy consumption of the multiply‐and‐accumulation operations during the inference phase compared to the conventional digital systems. This work paves the way for future autonomous systems at the edge.

Designing energy efficient systems capable to directly process signals at biological voltages remains a challenge. Here, the authors propose a bio-compatible memristor device based on protein-nanowire dielectric, harvested from the bacterium Geobactor sulfurreducens, working at biological voltages.

Though artificial sensory systems based on electronic devices have been realized, further transformation of data into spikes is required for neural network optimization. Here, based on NbO x Mott memristors, the authors report artificial spiking afferent nerves for accessing spiking systems and demonstrate spiking mechanoreceptor systems.

Sneak path current is a fundamental issue and a major roadblock to the wide application of memristor crossbar arrays. Traditional selectors such as transistors compromise the 2D scalability and 3D stack‐ability of the array, while emerging selectors with highly nonlinear current–voltage relations contradict the requirement of a linear current–voltage relation for efficient multiplication by directly using Ohm's law. Herein, the concept of a timing selector is proposed and demonstrated, which addresses the sneak path issue with a voltage‐dependent delay time of its transient switching behavior, while preserving a linear current–voltage relationship for computation. Crossbar arrays with silver‐based diffusive memristors as the timing selectors are built and the operation principle and operational windows are experimentally demonstrated. The timing selector enables large memristor crossbar arrays that can be used to solve large‐dimension real‐world problems in machine intelligence and neuromorphic computing.

Predicting functional outcomes after an Ischemic Stroke (IS) is highly valuable for patients and desirable for physicians. This facilitates physicians to set reasonable goals for patients and cooperate with patients and relatives effectively, and furthermore to reach common after-stroke care decisions for recovery and make exercise plans to facilitate rehabilitation. The objective of this research is to apply three current Deep Learning (DL) approaches for 6-month IS outcome predictions, using the openly accessible International Stroke Trial (IST) dataset. Furthermore, another objective of this research is to compare these DL approaches with machine learning (ML) for performing in clinical prediction. After comparing various ML methods (Deep Forest, Random Forest, Support Vector Machine, etc.) with current DL frameworks (CNN, LSTM, Resnet), the results show that DL doesn’t outperform ML significantly. DL methods and reporting used for analyzing structured medical data should be developed and improved.

The reservoir computing (RC) system, known for its ability to seamlessly integrate memory and computing functions, is considered as a promising solution to meet the high demands for time and energy-efficient computing in the current big data landscape, compared with traditional silicon-based computing systems that have a noticeable disadvantage of separate storage and computation. This review focuses on in-materio RC based on nanowire networks (NWs) from the perspective of materials, extending to reservoir devices and applications. The common methods used in preparing nanowires-based reservoirs, including the synthesis of nanowires and the construction of networks, are firstly systematically summarized. The physical principles of memristive and memcapacitive junctions are then explained. Afterwards, the dynamic characteristics of nanowires-based reservoirs and their computing capability, as well as the neuromorphic applications of NWs-based RC systems in recognition, classification, and forecasting tasks, are explicated in detail. Lastly, the current challenges and future opportunities facing NWs-based RC are highlighted, aiming to provide guidance for further research.

The development of tailings in concrete technology is not only conducive to the realization of the goal of reducing carbon emissions, but also conducive to the inhibition the occurrence of shortages of sand and gravel supplies. In this study, graphite tailings were used to replace sand in the range of 0~100%, and the mechanical mechanism of graphite tailings concrete was examined through compressive and flexural tests. The mechanical experimental results were evaluated and verified based on concrete macroscopic failure appearance, mesoscopic failure appearance, and physical characteristics of graphite tailings. The results revealed that the concrete strength increases first and then decreases with the increase of the graphite tailings content. Compared to GT00 (GT00 is a specimen with a graphite tailings content of 0%, and so on), GT10~GT60 exhibited better mechanical properties, of which 30% was recommended as the optimal replacement rate. The mechanical properties of GT10 and GT20 had an upward trend, and GT30 had low spalling, with aggregate fragmentation found on the fracture surface. GT30 showed the best resistance to bending and deformation. The mechanical properties of GT40~GT60 had a downward trend. When the graphite tailings content was higher than 70%, the interface defects of the aggregate matrix increased, thus making it easier for cracks to propagate along the interface. Furthermore, the mechanism of graphite tailings replacing sand verified the test results from different perspectives, which provides new analysis ideas for other tailings in environment-friendly concrete.

Inspired by competitive rules of the nature, competitive learning contributes to the specialization of the human brain and the general creativity of mankind. However, the construction of hardware competitive learning neural network still faces great challenges due to the lack of an accurate distance computation method and a self‐adaptive in situ training scheme. Herein, a fully memristive Euclidean distance (ED) engine based on analog multiply‐accumulate operation in a 32 × 32 TiN/TaO x /HfO x /TiN 1T1R array is demonstrated. The dual‐layer devices perform multilevel modulation under the target‐aware programming method with excellent read linearity in a dynamic range of 10–100 μS. The ED calculation is verified experimentally on a test board with an O(1) temporal complexity. Furthermore, in situ training and offline inference schemes for competitive learning, based on the ED engine, are developed and the simulated results show comparable success rates with those obtained by the CPU‐based software. Compared with a state‐of‐the‐art RTX6000 GPU (0.5 TOPS W−1), the energy efficiency of competitive learning models on ED engines can yield 100× improvements by utilizing optimized memristive devices.

Hepatocellular carcinoma (HCC) is one of the most prevalent malignant tumors worldwide. In view of the lack of early obvious clinical symptoms and related early diagnostic biomarkers with high specificity and sensitivity, most HCC patients are already at the advanced stages at the time of diagnosis, and most of them are accompanied by distant metastasis. Furthermore, the unsatisfactory effect of the follow-up palliative care contributes to the poor overall survival of HCC patients. Therefore, it is urgent to identify effective early diagnosis and prognostic biomarkers and to explore novel therapeutic approaches to improve the prognosis of HCC patients. Circular RNA (CircRNA), a class of plentiful, stable, and highly conserved ncRNA subgroup with the covalent closed loop, is dysregulated in HCC. Increasingly, emerging evidence have confirmed that dysregulated circRNAs can regulate gene expression at the transcriptional or post-transcriptional level, mediating various malignant biological behaviors of HCC cells, including proliferation, invasion, metastasis, immune escape, stemness, and drug resistance, etc.; meanwhile, they are regarded as potential biomarkers for early diagnosis and prognostic evaluation of HCC. This article reviews the research progress of circRNAs in HCC, expounding the potential molecular mechanisms of dysregulated circRNAs in the carcinogenesis and development of HCC, and discusses those application prospects in the diagnosis and prognosis of HCC.

Long non-coding RNA (lncRNA), a subgroup of ncRNA with a length of more than 200 nt without protein coding function, has been recognized by the academia for its mediating effects of dysregulated expression on the tumorigenesis and development of a variety of tumors. LncRNA DiGeorge syndrome critical region gene 5 (DGCR5), originally found to induce DiGeorge syndrome, has been confirmed to be extremely dysregulated in multiple tumors, which mediates the malignant phenotypes of hepatocellular carcinoma, pancreatic cancer, lung cancer, etc. through the regulation of Wnt/β-catenin, MEK/ERK1/2 and other cancerous signaling pathways as a molecular sponge. Researches on the cancerous derivation-related pathways involved in DGCR5 can provide potential molecular intervention targets for tumor precision treatment. Moreover, liquid biopsy based on the detection of DGCR5 in body fluids is also expected to provide a non-invasive evaluation method for the early diagnosis and prognostic evaluation of malignant tumors.

The emergence of memristors with potential applications in data storage and artificial intelligence has attracted wide attentions. Memristors are assembled in crossbar arrays with data bits encoded by the resistance of individual cells. Despite the proposed high density and excellent scalability, the sneak‐path current causing cross interference impedes their practical applications. Therefore, developing novel architectures to mitigate sneak‐path current and improve efficiency, reliability, and stability may benefit next‐generation storage‐class memory (SCM). Moreover, conventional digital computers face the von‐Neumann bottleneck and the slowdown of transistors’ scaling, imposing a big challenge to hardware artificial intelligence. Memristive crossbar features colocation of memory and processing units, as well as superior scalability, making it a promising candidate for hardware accelerating machine learning and neuromorphic computing. Herein, first, crossbar architecture is introduced. Then, for storage, the origin of sneak‐path current is reviewed and techniques to mitigate this issue from the angle of materials and circuits are discussed. Computing wise, the applications of memristive crossbars in both machine learning and neuromorphic computing are surveyed, focusing on the structure of unit cells, the network topology, and the learning types. Finally, a perspective on future engineering and applications of memristive crossbars is discussed.

The rapid development of neural networks has led to tremendous applications in image segmentation, speech recognition, and medical image diagnosis, etc. Among various hardware implementations of neural networks, silicon photonics is considered one of the most promising approaches due to its CMOS compatibility, accessible integration platforms, mature fabrication techniques, and abundant optical components. In addition, neuromorphic computing based on silicon photonics can provide massively parallel processing and high-speed operations with low power consumption, thus enabling further exploration of neural networks. Here, we focused on the development of neuromorphic computing based on silicon photonics, introducing this field from the perspective of electronic–photonic co-design and presenting the architecture and algorithm theory. Finally, we discussed the prospects and challenges of neuromorphic silicon photonics.

Hepatocellular carcinoma (HCC) has become a challenging disease in the world today. Due to the limitations on the current diagnosis and treatment as well as its high metastatic ability and high recurrence rate, HCC gradually becomes the second deadliest tumor. Exosomes are one of the types of cell-derived vesicles and can carry intracellular materials such as genetic materials, lipids, and proteins. In recent years, it has been verified that exosomes are linked to numerous physiological and pathological processes, including HCC. However, how exosomes affect HCC progression remains largely unknown. In this review, the exosome-mediated cellular material transfer between cells of different types in the HCC microenvironment and their effects on the behaviors and functions of recipient cells are studied. Furthermore, we also addressed the underlying molecular mechanisms. We believe that new light on the diagnosis of this cancer as well as its treatment strategies will be shed after a collation of literature in this area.

With the slowdown of Moore’s law, many emerging electronic devices and computing architectures have been proposed to sustain the performance advancement of computing. Among them, the Ising machine is a non-von-Neumann solver that has received wide attention in recent years. It is capable of solving intractable combinatorial optimization (CO) problems, which are difficult to be solve using conventional digital computers. In fact, many CO problems can be mapped to finding the corresponding ground states of Ising model. At present, Ising machine prototypes based on different physical principles, such as emerging memristive oscillators, have been demonstrated, among which the Ising Hamiltonian solver based on the coupled oscillator network simultaneously holds the advantages of room-temperature operation, compact footprint, low power consumption, and fast speed to solution. This paper comprehensively surveys the recent developments in this important field, including the types of oscillators, the implementation principle of the Ising model, and the solver’s performance. Finally, methods to further improve the performance have also been suggested.

Abstract Background The solute carrier (SLC) 7 family genes comprise 14 members and function as cationic amino acid/glycoprotein transporters in many cells, they are essential for the maintenance of amino acid nutrition and survival of tumor cells. This study was conducted to analyze the associations of SLC7 family gene expression with mortality in papillary thyroid carcinoma (PTC). Methods Clinical features, somatic mutations, and SLC7 family gene expression data were downloaded from The Cancer Genome Atlas database. Linear regression model analysis was performed to analyze the correlations between SLC7 family gene expression and clinicopathologic features. Kaplan-Meier survival and logistic regression analyses were performed to characterize the associations between gene expression and patients’ overall survival. Results Patient mortality was negatively associated with age and tumor size but positively increased cancer stage and absence of thyroiditis in PTC patients. Kaplan-Meier survival analysis indicated that patients with high SLC7A3, SLC7A5, and SLC7A11 expression levels exhibited poorer survival than those with low SLC7A3, SLC7A5, and SLC7A11 expression levels (P

Cancer stem cells (CSCs) contribute to the cancer initiation, metastasis and drug resistance in non-small cell lung cancer (NSCLC). Herein, we identified a miR-221/222 cluster as a novel regulator of CSCs in NSCLC. Targeted overexpression or knockdown of miR-221/222 in NSCLC cells revealed the essential roles of miR-221/222 in regulation of lung cancer cell proliferation, mammosphere formation, subpopulation of CD133+ CSCs and the expression of stemness genes including OCT4, NANOG and h-TERT. The in vivo animal study showed that overexpression of miR-221/222 significantly enhanced the capacity of lung cancer cells to develop tumor and grow faster, indicating the importance of miR-221/222 in tumorigenesis and tumor growth. Mechanistically, Reck was found to be a key direct target gene of miR-221/222 in NSCLC. Overexpression of miR-221/222 significantly suppressed Reck expression, activated Notch1 signaling and increased the level of NICD. As an activated form of Notch1, NICD leads to enhanced stemness in NSCLC cells. In addition, knockdown of Reck by siRNA not only mimicked miR-221/222 effects, but also demonstrated involvement of Reck in the miR-221/222-induced activation of Notch1 signaling, verifying the essential roles of the miR-221/222-Reck-Notch1 axis in regulating stemness of NSCLC cells. These findings uncover a novel mechanism by which lung CSCs are significantly manipulated by miR-221/222, and provide a potential therapeutic target for the treatment of NSCLC.

Exosomes are small extracellular vesicles secreted by most somatic cells, which can carry a variety of biologically active substances to participate in intercellular communication and regulate the pathophysiological process of recipient cells. Recent studies have confirmed that non-coding RNAs (ncRNAs) carried by tumor cell/non-tumor cell-derived exosomes have the function of regulating the cancerous derivation of target cells and remodeling the tumor microenvironment (TME). In addition, due to the unique low immunogenicity and high stability, exosomes can be used as natural vehicles for the delivery of therapeutic ncRNAs in vivo. This article aims to review the potential regulatory mechanism and the therapeutic value of exosomal ncRNAs in hepatocellular carcinoma (HCC), in order to provide promising targets for early diagnosis and precise therapy of HCC.

As impacted by environmental concerns and the demand for high-strength concrete, novel ideas of the development of eco-efficient, steel fiber-reinforced concrete have been proposed. In this study, the aim is to develop a type of eco-efficient steel fiber-reinforced concrete with graphite tailings and steel fiber. Steel fibers act as a type of concrete toughening material, and graphite tailings serve as a partially alternated aggregate to sand. Mechanical properties exhibited by the concrete are assessed based on different volume fractions of graphite tailings (i.e., 0%, 10% and 20%). The concrete mixture proportion is determined in accordance with the theory of particle densely packing, and the concrete mechanical properties are more specifically studied by performing compressive and flexural tests. As indicated by the results, the maximal mixed bulk density of graphite tailings at different grades is greater than that of sand, so concrete with graphite tailings exhibits higher compressive strength. For the content of graphite tailings, the addition of graphite tailings impacts the interfacial adhesion between aggregates and cementing matrix, thereby inhibiting bifurcation and convergence of cracks. However, excessive mixing of graphite tailings would decrease the specimen’s effective water-to-binder ratio (W/B), thereby adversely affecting the internal structure of the concrete. The amount of graphite tailings impacts the distribution of steel fibers at the concrete interface, which could be the most conducive to the distribution of steel fibers under the graphite tailings’ content of 10%. This study demonstrates that graphite tailings and steel fibers are feasible to prepare eco-efficient, steel fiber-reinforced concrete.

In this study, the combined effects of different amounts of graphite tailings (GT) as a substitute for sand and three cure methods on the early-age mechanical and material performances of cement mortar have been investigated. The results can be concluded that graphite tailings and hot-water cure method can effectively improve the early strength of cement-based materials and facilitate the formation of high-density calcium silicate hydrate (C-S-H), more Ca(OH)2 (CH), and other hydration products. Simultaneously, the pore structure distribution in the cement mortar is improved to be finer. Finally, this paper explains the mechanism of GT and hot-water curing (HWC) method on early-age performances of the cement mortar through the graphic model and provides the basis for selecting the GT and HWC method on improving early performance requirements for special engineering or rapid repair engineering through the proposed comprehensive quantitative calculation model.

Cisplatin has been widely used in the treatment of a various types of cancers including triple-negative breast cancer (TNBC) by damaging DNA and inducing apoptosis. However, its anti-cancer effects are often limited due to chemo-resistance, which is one of the main reasons causing cancer relapse and metastasis. To overcome resistance, cisplatin is often used in combination with other drugs or molecules. Our study found that the targeted inhibition of miR-221/222 in MDA-MB-231 cells promoted cisplatin-induced cell apoptosis, and increased the cell sensitivity to cisplatin in vitro. Much higher expression levels of miR-221/222 were detected in the cisplatin-resistant MDA-MB-231 cells and in cisplatin-resistant breast cancer patients. The combination chemotherapy of cisplatin with anti-miR-221/222 showed much higher efficiency in suppressing tumor growth in the mice transplanted with MDA-MB-231 cells. In addition, anti-miR-221 and anti-miR-222 showed synergetic effects on improving sensitivity to cisplatin in MDA-MB-231 cells. Suppression of SOCS1-STAT3-Bcl-2 pathway and activation of p53-Pten signaling both contribute to anti-miR-221/222-induced sensitivity to cisplatin in MDA-MB-231 cells. These findings suggest the potential of a novel approach for the combination chemotherapy of cisplatin with small non-coding RNA in treatment of human TNBC.

Though memristors can potentially emulate neuron and synapse functionality, useful signal energy is lost to Joule heating. Here, the authors demonstrate neuro-transistors with a pseudo-memcapacitive gate that actively process signals via energy-efficient capacitively-coupled neural networks.

Memristor-based neural networks hold promise for neuromorphic computing, yet large-scale experimental execution remains difficult. Here, Xia et al. create a multi-layer memristor neural network with in-situ machine learning and achieve competitive image classification accuracy on a standard dataset.