Pluggable multitask diffractive neural networks based on cascaded metasurfaces

Optical neural networks have significant advantages in terms of power consumption,parallelism,and high computing speed,which has intrigued extensive attention in both academic and engineering communities.It has been considered as one of the powerful tools in promoting the fields of imaging processing and object recognition.However,the existing optical system architecture cannot be reconstructed to the realization of multi-functional artificial intelligence systems simultaneously.To push the development of this issue,we propose the pluggable diffractive neural networks(P-DNN),a general paradigm resorting to the cascaded metasurfaces,which can be applied to recognize various tasks by switching internal plug-ins.As the proof-of-principle,the recognition functions of six types of handwritten digits and six types of fashions are numerical simulated and experimental demonstrated at near-infrared regimes.Encouragingly,the proposed paradigm not only improves the flexibility of the optical neural networks but paves the new route for achieving high-speed,low-power and versatile artificial intelligence systems.

Emerging strategies for nerve repair and regeneration in ischemic stroke:neural stem cell therapy

Ischemic stroke is a major cause of mortality and disability worldwide,with limited treatment options available in clinical practice.The emergence of stem cell therapy has provided new hope to the field of stroke treatment via the restoration of brain neuron function.Exogenous neural stem cells are beneficial not only in cell replacement but also through the bystander effect.Neural stem cells regulate multiple physiological responses,including nerve repair,endogenous regeneration,immune function,and blood-brain barrier permeability,through the secretion of bioactive substances,including extracellular vesicles/exosomes.However,due to the complex microenvironment of ischemic cerebrovascular events and the low survival rate of neural stem cells following transplantation,limitations in the treatment effect remain unresolved.In this paper,we provide a detailed summary of the potential mechanisms of neural stem cell therapy for the treatment of ischemic stroke,review current neural stem cell therapeutic strategies and clinical trial results,and summarize the latest advancements in neural stem cell engineering to improve the survival rate of neural stem cells.We hope that this review could help provide insight into the therapeutic potential of neural stem cells and guide future scientific endeavors on neural stem cells.

Recent advances in the application of MXenes for neural tissue engineering and regeneration

Transition metal carbides and nitrides(MXenes)are crystal nanomaterials with a number of surface functional groups such as fluorine,hydroxyl,and oxygen,which can be used as carriers for proteins and drugs.MXenes have excellent biocompatibility,electrical conductivity,surface hydrophilicity,mechanical properties and easy surface modification.However,at present,the stability of most MXenes needs to be improved,and more synthesis methods need to be explored.MXenes are good substrates for nerve cell regeneration and nerve reconstruction,which have broad application prospects in the repair of nervous system injury.Regarding the application of MXenes in neuroscience,mainly at the cellular level,the long-term in vivo biosafety and effects also need to be further explored.This review focuses on the progress of using MXenes in nerve regeneration over the last few years;discussing preparation of MXenes and their biocompatibility with different cells as well as the regulation by MXenes of nerve cell regeneration in two-dimensional and three-dimensional environments in vitro.MXenes have great potential in regulating the proliferation,differentiation,and maturation of nerve cells and in promoting regeneration and recovery after nerve injury.In addition,this review also presents the main challenges during optimization processes,such as the preparation of stable MXenes and long-term in vivo biosafety,and further discusses future directions in neural tissue engineering.

Autophagy in neural stem cells and glia for brain health and diseases

Autophagy is a multifaceted cellular process that not only maintains the homeostatic and adaptive responses of the brain but is also dynamically involved in the regulation of neural cell generation,maturation,and survival.Autophagy facilities the utilization of energy and the microenvironment for developing neural stem cells.Autophagy arbitrates structural and functional remodeling during the cell differentiation process.Autophagy also plays an indispensable role in the maintenance of stemness and homeostasis in neural stem cells during essential brain physiology and also in the instigation and progression of diseases.Only recently,studies have begun to shed light on autophagy regulation in glia(microglia,astrocyte,and oligodendrocyte)in the brain.Glial cells have attained relatively less consideration despite their unquestioned influence on various aspects of neural development,synaptic function,brain metabolism,cellular debris clearing,and restoration of damaged or injured tissues.Thus,this review composes pertinent information regarding the involvement of autophagy in neural stem cells and glial regulation and the role of this connexion in normal brain functions,neurodevelopmental disorders,and neurodegenerative diseases.This review will provide insight into establishing a concrete strategic approach for investigating pathological mechanisms and developing therapies for brain diseases.

Implantable probe with integrated reference electrode for in situ neural signal and calcium ion monitoring

Monitoring the electrophysiology activity of neurons and blood calcium signals can enable a better understanding of disease-related neural system circuits.However,currently,in situ calcium ion monitoring tools are scarce and exhibit low integration and limited sensitivity.In this letter,we propose an implantable probe with an integrated in situ Ag/AgCl reference electrode(ISA/ARE)that can monitor action potential(AP)and Ca^(2+) concentrations.

Metabolic and proteostatic differences in quiescent and active neural stem cells

Adult neural stem cells are neurogenesis progenitor cells that play an important role in neurogenesis.Therefore,neural regeneration may be a promising target for treatment of many neurological illnesses.The regenerative capacity of adult neural stem cells can be chara cterized by two states:quiescent and active.Quiescent adult neural stem cells are more stable and guarantee the quantity and quality of the adult neural stem cell pool.Active adult neural stem cells are chara cterized by rapid proliferation and differentiation into neurons which allow for integration into neural circuits.This review focuses on diffe rences between quiescent and active adult neural stem cells in nutrition metabolism and protein homeostasis.Furthermore,we discuss the physiological significance and underlying advantages of these diffe rences.Due to the limited number of adult neural stem cells studies,we refe rred to studies of embryonic adult neural stem cells or non-mammalian adult neural stem cells to evaluate specific mechanisms.

Activation Redistribution Based Hybrid Asymmetric Quantization Method of Neural Networks

The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedded devices.In order to reduce the complexity and overhead of deploying neural networks on Integeronly hardware,most current quantization methods use a symmetric quantization mapping strategy to quantize a floating-point neural network into an integer network.However,although symmetric quantization has the advantage of easier implementation,it is sub-optimal for cases where the range could be skewed and not symmetric.This often comes at the cost of lower accuracy.This paper proposed an activation redistribution-based hybrid asymmetric quantizationmethod for neural networks.The proposedmethod takes data distribution into consideration and can resolve the contradiction between the quantization accuracy and the ease of implementation,balance the trade-off between clipping range and quantization resolution,and thus improve the accuracy of the quantized neural network.The experimental results indicate that the accuracy of the proposed method is 2.02%and 5.52%higher than the traditional symmetric quantization method for classification and detection tasks,respectively.The proposed method paves the way for computationally intensive neural network models to be deployed on devices with limited computing resources.Codes will be available on https://github.com/ycjcy/Hybrid-Asymmetric-Quantization.

Neural stem cells promote neuroplasticity: a promising therapeutic strategy for the treatment of Alzheimer’s disease

Recent studies have demonstrated that neuroplasticity,such as synaptic plasticity and neurogenesis,exists throughout the normal lifespan but declines with age and is significantly impaired in individuals with Alzheimer’s disease.Hence,promoting neuroplasticity may represent an effective strategy with which Alzheimer’s disease can be alleviated.Due to their significant ability to self-renew,differentiate,and migrate,neural stem cells play an essential role in reversing synaptic and neuronal damage,reducing the pathology of Alzheimer’s disease,including amyloid-β,tau protein,and neuroinflammation,and secreting neurotrophic factors and growth factors that are related to plasticity.These events can promote synaptic plasticity and neurogenesis to repair the microenvironment of the mammalian brain.Consequently,neural stem cells are considered to represent a potential regenerative therapy with which to improve Alzheimer’s disease and other neurodegenerative diseases.In this review,we discuss how neural stem cells regulate neuroplasticity and optimize their effects to enhance their potential for treating Alzheimer’s disease in the clinic.

Biomaterials and tissue engineering in traumatic brain injury:novel perspectives on promoting neural regeneration

Traumatic brain injury is a serious medical condition that can be attributed to falls, motor vehicle accidents, sports injuries and acts of violence, causing a series of neural injuries and neuropsychiatric symptoms. However, limited accessibility to the injury sites, complicated histological and anatomical structure, intricate cellular and extracellular milieu, lack of regenerative capacity in the native cells, vast variety of damage routes, and the insufficient time available for treatment have restricted the widespread application of several therapeutic methods in cases of central nervous system injury. Tissue engineering and regenerative medicine have emerged as innovative approaches in the field of nerve regeneration. By combining biomaterials, stem cells, and growth factors, these approaches have provided a platform for developing effective treatments for neural injuries, which can offer the potential to restore neural function, improve patient outcomes, and reduce the need for drugs and invasive surgical procedures. Biomaterials have shown advantages in promoting neural development, inhibiting glial scar formation, and providing a suitable biomimetic neural microenvironment, which makes their application promising in the field of neural regeneration. For instance, bioactive scaffolds loaded with stem cells can provide a biocompatible and biodegradable milieu. Furthermore, stem cells-derived exosomes combine the advantages of stem cells, avoid the risk of immune rejection, cooperate with biomaterials to enhance their biological functions, and exert stable functions, thereby inducing angiogenesis and neural regeneration in patients with traumatic brain injury and promoting the recovery of brain function. Unfortunately, biomaterials have shown positive effects in the laboratory, but when similar materials are used in clinical studies of human central nervous system regeneration, their efficacy is unsatisfactory. Here, we review the characteristics and properties of various bioactive materials, followed by the introduction of applications based on biochemistry and cell molecules, and discuss the emerging role of biomaterials in promoting neural regeneration. Further, we summarize the adaptive biomaterials infused with exosomes produced from stem cells and stem cells themselves for the treatment of traumatic brain injury. Finally, we present the main limitations of biomaterials for the treatment of traumatic brain injury and offer insights into their future potential.

Transplantation of fibrin-thrombin encapsulated human induced neural stem cells promotes functional recovery of spinal cord injury rats through modulation of the microenvironment

Recent studies have mostly focused on engraftment of cells at the lesioned spinal cord,with the expectation that differentiated neurons facilitate recovery.Only a few studies have attempted to use transplanted cells and/or biomaterials as major modulators of the spinal cord injury microenvironment.Here,we aimed to investigate the role of microenvironment modulation by cell graft on functional recovery after spinal cord injury.Induced neural stem cells reprogrammed from human peripheral blood mononuclear cells,and/or thrombin plus fibrinogen,were transplanted into the lesion site of an immunosuppressed rat spinal cord injury model.Basso,Beattie and Bresnahan score,electrophysiological function,and immunofluorescence/histological analyses showed that transplantation facilitates motor and electrophysiological function,reduces lesion volume,and promotes axonal neurofilament expression at the lesion core.Examination of the graft and niche components revealed that although the graft only survived for a relatively short period(up to 15 days),it still had a crucial impact on the microenvironment.Altogether,induced neural stem cells and human fibrin reduced the number of infiltrated immune cells,biased microglia towards a regenerative M2 phenotype,and changed the cytokine expression profile at the lesion site.Graft-induced changes of the microenvironment during the acute and subacute stages might have disrupted the inflammatory cascade chain reactions,which may have exerted a long-term impact on the functional recovery of spinal cord injury rats.

Multi-Scale-Matching neural networks for thin plate bending problem

Physics-informed neural networks are a useful machine learning method for solving differential equations,but encounter challenges in effectively learning thin boundary layers within singular perturbation problems.To resolve this issue,multi-scale-matching neural networks are proposed to solve the singular perturbation problems.Inspired by matched asymptotic expansions,the solution is decomposed into inner solutions for small scales and outer solutions for large scales,corresponding to boundary layers and outer regions,respectively.Moreover,to conform neural networks,we introduce exponential stretched variables in the boundary layers to avoid semiinfinite region problems.Numerical results for the thin plate problem validate the proposed method.

Long non-coding RNA H19 regulates neurogenesis of induced neural stem cells in a mouse model of closed head injury

Stem cell-based therapies have been proposed as a potential treatment for neural regeneration following closed head injury.We previously reported that induced neural stem cells exert beneficial effects on neural regeneration via cell replacement.However,the neural regeneration efficiency of induced neural stem cells remains limited.In this study,we explored differentially expressed genes and long non-coding RNAs to clarify the mechanism underlying the neurogenesis of induced neural stem cells.We found that H19 was the most downregulated neurogenesis-associated lnc RNA in induced neural stem cells compared with induced pluripotent stem cells.Additionally,we demonstrated that H19 levels in induced neural stem cells were markedly lower than those in induced pluripotent stem cells and were substantially higher than those in induced neural stem cell-derived neurons.We predicted the target genes of H19 and discovered that H19 directly interacts with mi R-325-3p,which directly interacts with Ctbp2 in induced pluripotent stem cells and induced neural stem cells.Silencing H19 or Ctbp2 impaired induced neural stem cell proliferation,and mi R-325-3p suppression restored the effect of H19 inhibition but not the effect of Ctbp2 inhibition.Furthermore,H19 silencing substantially promoted the neural differentiation of induced neural stem cells and did not induce apoptosis of induced neural stem cells.Notably,silencing H19 in induced neural stem cell grafts markedly accelerated the neurological recovery of closed head injury mice.Our results reveal that H19 regulates the neurogenesis of induced neural stem cells.H19 inhibition may promote the neural differentiation of induced neural stem cells,which is closely associated with neurological recovery following closed head injury.

Evolutionary Neural Architecture Search and Its Applications in Healthcare

Most of the neural network architectures are based on human experience,which requires a long and tedious trial-and-error process.Neural architecture search(NAS)attempts to detect effective architectures without human intervention.Evolutionary algorithms(EAs)for NAS can find better solutions than human-designed architectures by exploring a large search space for possible architectures.Using multiobjective EAs for NAS,optimal neural architectures that meet various performance criteria can be explored and discovered efficiently.Furthermore,hardware-accelerated NAS methods can improve the efficiency of the NAS.While existing reviews have mainly focused on different strategies to complete NAS,a few studies have explored the use of EAs for NAS.In this paper,we summarize and explore the use of EAs for NAS,as well as large-scale multiobjective optimization strategies and hardware-accelerated NAS methods.NAS performs well in healthcare applications,such as medical image analysis,classification of disease diagnosis,and health monitoring.EAs for NAS can automate the search process and optimize multiple objectives simultaneously in a given healthcare task.Deep neural network has been successfully used in healthcare,but it lacks interpretability.Medical data is highly sensitive,and privacy leaks are frequently reported in the healthcare industry.To solve these problems,in healthcare,we propose an interpretable neuroevolution framework based on federated learning to address search efficiency and privacy protection.Moreover,we also point out future research directions for evolutionary NAS.Overall,for researchers who want to use EAs to optimize NNs in healthcare,we analyze the advantages and disadvantages of doing so to provide detailed guidance,and propose an interpretable privacy-preserving framework for healthcare applications.

Multi-scale physics-informed neural networks for solving high Reynolds number boundary layer flows based on matched asymptotic expansions

Multi-scale system remains a classical scientific problem in fluid dynamics,biology,etc.In the present study,a scheme of multi-scale Physics-informed neural networks is proposed to solve the boundary layer flow at high Reynolds numbers without any data.The flow is divided into several regions with different scales based on Prandtl's boundary theory.Different regions are solved with governing equations in different scales.The method of matched asymptotic expansions is used to make the flow field continuously.A flow on a semi infinite flat plate at a high Reynolds number is considered a multi-scale problem because the boundary layer scale is much smaller than the outer flow scale.The results are compared with the reference numerical solutions,which show that the msPINNs can solve the multi-scale problem of the boundary layer in high Reynolds number flows.This scheme can be developed for more multi-scale problems in the future.

A data-driven model of drop size prediction based on artificial neural networks using small-scale data sets

An artificial neural network(ANN)method is introduced to predict drop size in two kinds of pulsed columns with small-scale data sets.After training,the deviation between calculate and experimental results are 3.8%and 9.3%,respectively.Through ANN model,the influence of interfacial tension and pulsation intensity on the droplet diameter has been developed.Droplet size gradually increases with the increase of interfacial tension,and decreases with the increase of pulse intensity.It can be seen that the accuracy of ANN model in predicting droplet size outside the training set range is reach the same level as the accuracy of correlation obtained based on experiments within this range.For two kinds of columns,the drop size prediction deviations of ANN model are 9.6%and 18.5%and the deviations in correlations are 11%and 15%.

Non-crossing Quantile Regression Neural Network as a Calibration Tool for Ensemble Weather Forecasts

Despite the maturity of ensemble numerical weather prediction(NWP),the resulting forecasts are still,more often than not,under-dispersed.As such,forecast calibration tools have become popular.Among those tools,quantile regression(QR)is highly competitive in terms of both flexibility and predictive performance.Nevertheless,a long-standing problem of QR is quantile crossing,which greatly limits the interpretability of QR-calibrated forecasts.On this point,this study proposes a non-crossing quantile regression neural network(NCQRNN),for calibrating ensemble NWP forecasts into a set of reliable quantile forecasts without crossing.The overarching design principle of NCQRNN is to add on top of the conventional QRNN structure another hidden layer,which imposes a non-decreasing mapping between the combined output from nodes of the last hidden layer to the nodes of the output layer,through a triangular weight matrix with positive entries.The empirical part of the work considers a solar irradiance case study,in which four years of ensemble irradiance forecasts at seven locations,issued by the European Centre for Medium-Range Weather Forecasts,are calibrated via NCQRNN,as well as via an eclectic mix of benchmarking models,ranging from the naïve climatology to the state-of-the-art deep-learning and other non-crossing models.Formal and stringent forecast verification suggests that the forecasts post-processed via NCQRNN attain the maximum sharpness subject to calibration,amongst all competitors.Furthermore,the proposed conception to resolve quantile crossing is remarkably simple yet general,and thus has broad applicability as it can be integrated with many shallow-and deep-learning-based neural networks.

Development of a convolutional neural network based geomechanical upscaling technique for heterogeneous geological reservoir

Geomechanical assessment using coupled reservoir-geomechanical simulation is becoming increasingly important for analyzing the potential geomechanical risks in subsurface geological developments.However,a robust and efficient geomechanical upscaling technique for heterogeneous geological reservoirs is lacking to advance the applications of three-dimensional(3D)reservoir-scale geomechanical simulation considering detailed geological heterogeneities.Here,we develop convolutional neural network(CNN)proxies that reproduce the anisotropic nonlinear geomechanical response caused by lithological heterogeneity,and compute upscaled geomechanical properties from CNN proxies.The CNN proxies are trained using a large dataset of randomly generated spatially correlated sand-shale realizations as inputs and simulation results of their macroscopic geomechanical response as outputs.The trained CNN models can provide the upscaled shear strength(R^(2)>0.949),stress-strain behavior(R^(2)>0.925),and volumetric strain changes(R^(2)>0.958)that highly agree with the numerical simulation results while saving over two orders of magnitude of computational time.This is a major advantage in computing the upscaled geomechanical properties directly from geological realizations without the need to perform local numerical simulations to obtain the geomechanical response.The proposed CNN proxybased upscaling technique has the ability to(1)bridge the gap between the fine-scale geocellular models considering geological uncertainties and computationally efficient geomechanical models used to assess the geomechanical risks of large-scale subsurface development,and(2)improve the efficiency of numerical upscaling techniques that rely on local numerical simulations,leading to significantly increased computational time for uncertainty quantification using numerous geological realizations.

Hybrid model for BOF oxygen blowing time prediction based on oxygen balance mechanism and deep neural network

The amount of oxygen blown into the converter is one of the key parameters for the control of the converter blowing process,which directly affects the tap-to-tap time of converter. In this study, a hybrid model based on oxygen balance mechanism (OBM) and deep neural network (DNN) was established for predicting oxygen blowing time in converter. A three-step method was utilized in the hybrid model. First, the oxygen consumption volume was predicted by the OBM model and DNN model, respectively. Second, a more accurate oxygen consumption volume was obtained by integrating the OBM model and DNN model. Finally, the converter oxygen blowing time was calculated according to the oxygen consumption volume and the oxygen supply intensity of each heat. The proposed hybrid model was verified using the actual data collected from an integrated steel plant in China, and compared with multiple linear regression model, OBM model, and neural network model including extreme learning machine, back propagation neural network, and DNN. The test results indicate that the hybrid model with a network structure of 3 hidden layer layers, 32-16-8 neurons per hidden layer, and 0.1 learning rate has the best prediction accuracy and stronger generalization ability compared with other models. The predicted hit ratio of oxygen consumption volume within the error±300 m^(3)is 96.67%;determination coefficient (R^(2)) and root mean square error (RMSE) are0.6984 and 150.03 m^(3), respectively. The oxygen blow time prediction hit ratio within the error±0.6 min is 89.50%;R2and RMSE are0.9486 and 0.3592 min, respectively. As a result, the proposed model can effectively predict the oxygen consumption volume and oxygen blowing time in the converter.

Neural-Polar码:一种基于深度学习的新型信道编码方案

为应对新型移动通信系统智能性的需求以及在难以进行人工建模的复杂信道环境下进行可靠通信的问题,基于Polar码的编译码递归结构提出一种新型神经网络信道编码方案,即Neural-Polar码。该方案利用神经网络将Polar码编译码递归结构中父、子节点间的线性映射变成非线性映射,引入快速连续抵消(successive cancellation, SC)译码的思想,解决在完全二叉树上构建Neural-Polar码造成网络结构过大的问题。仿真实验表明,Neural-Polar码可以获得优于经典SC译码算法的误码率(bit error rate, BER)和误块率(block error rate, BLER)性能,对网络的联合训练使得Neural-Polar码能够自动学习信道特性,具有更好的信道适应性和鲁棒性。Neural-Polar码将传统的对复杂信道进行人工建模分析的难题交给机器,充分体现出其编译码的智能性。

2015年3月特大磁暴期间中国区域电离层TEC NeuralProphet预报模型研究

延迟是全球卫星导航定位中重要的误差源之一,提高电离层TEC建模和预报精度对改善卫星导航定位精度至关重要.本文构建了以太阳辐射通量指数F_(10.7)、地磁活动指数Dst、地理坐标和中国科学院(Chinese Academy of Sciences,CAS)GIM数据为输入参数的NeuralProphet神经网络模型(NP模型),实现在2015年3月特大磁暴期中国区域电离层TEC短期预报.为验证NP模型的预报精度,本文同时构建了长短期记忆神经网络(Long Short-term Memory Neural Network,LSTM)模型进行对比分析.结果统计分析表明,NP模型在磁暴期(2015年DOY076-078)TEC预报值RMSE和RD分别为0.83 TECU和3.13%,绝对和相对精度较LSTM模型分别提高1.49 TECU和10.25%;且NP模型RMSE优于1.5 TECU的比例达97.24%,远高于LSTM模型.NP模型预报值与CAS具有较好一致性和无偏性,偏差均值仅为-0.01 TECU,而LSTM模型预报值的均值偏大,偏差均值为1.49 TECU.从低纬到中纬度的三个纬度带内,NP模型RMSE分别为1.12、0.83和0.44 TECU,精度比LSTM模型提高1.94、1.56和1.23 TECU.整体上,在磁暴期NP模型预报性能明显优于LSTM模型,能够精细描述中国区域电离层TEC时空变化.