RepBoTNet-CESA:An Alzheimer’s Disease Computer Aided Diagnosis Method Using Structural Reparameterization BoTNet and Cubic Embedding Self Attention

Various deep learning models have been proposed for the accurate assisted diagnosis of early-stage Alzheimer’s disease(AD).Most studies predominantly employ Convolutional Neural Networks(CNNs),which focus solely on local features,thus encountering difficulties in handling global features.In contrast to natural images,Structural Magnetic Resonance Imaging(sMRI)images exhibit a higher number of channel dimensions.However,during the Position Embedding stage ofMulti Head Self Attention(MHSA),the coded information related to the channel dimension is disregarded.To tackle these issues,we propose theRepBoTNet-CESA network,an advanced AD-aided diagnostic model that is capable of learning local and global features simultaneously.It combines the advantages of CNN networks in capturing local information and Transformer networks in integrating global information,reducing computational costs while achieving excellent classification performance.Moreover,it uses the Cubic Embedding Self Attention(CESA)proposed in this paper to incorporate the channel code information,enhancing the classification performance within the Transformer structure.Finally,the RepBoTNet-CESA performs well in various AD-aided diagnosis tasks,with an accuracy of 96.58%,precision of 97.26%,and recall of 96.23%in the AD/NC task;an accuracy of 92.75%,precision of 92.84%,and recall of 93.18%in the EMCI/NC task;and an accuracy of 80.97%,precision of 83.86%,and recall of 80.91%in the AD/EMCI/LMCI/NC task.This demonstrates that RepBoTNet-CESA delivers outstanding outcomes in various AD-aided diagnostic tasks.Furthermore,our study has shown that MHSA exhibits superior performance compared to conventional attention mechanisms in enhancing ResNet performance.Besides,the Deeper RepBoTNet-CESA network fails to make further progress in AD-aided diagnostic tasks.

基于selffer net构建的人工智能模型在风机叶根螺栓失效预测场景下的应用

针对风机叶根螺栓失效问题,文章提出了一种基于selffer net构建的叠加树人工智能模型。该模型集成了回溯训练层,叠加树模型,遮蔽特征模型,以及LightGBM模型,实现了对风机叶根螺栓失效概率的预测。经过实验验证,该模型在预测风机叶根螺栓失效概率方面具有较高的准确性和稳定性,为风机叶根螺栓失效预测场景提供了有效的解决方案。

基于选择性自校正卷积U-Net的肺部X射线图像肺实质分割

针对U-Net分割算法无法提取多尺度特征、易受到伪影和噪声干扰而导致在肺部X射线图像中肺实质分割不精确的问题,提出一种基于选择性自校正卷积的U-Net改进算法。改进后的U-Net算法将普通卷积模块替换为选择性自校正卷积模块,该模块采用多分支结构提取多尺度特征信息,使用Sigmoid函数和Softmax函数对多尺度特征信息进行选择性校正,使校正后的特征信息聚焦于肺实质区域,输出特征更加具有针对性。实验表明,该方法对骰子系数、交并比、F_(1)评分结果以及对肺实质分割结果都有一定程度的提升。

Short‐term and long‐term memory self‐attention network for segmentation of tumours in 3D medical images

Tumour segmentation in medical images(especially 3D tumour segmentation)is highly challenging due to the possible similarity between tumours and adjacent tissues,occurrence of multiple tumours and variable tumour shapes and sizes.The popular deep learning‐based segmentation algorithms generally rely on the convolutional neural network(CNN)and Transformer.The former cannot extract the global image features effectively while the latter lacks the inductive bias and involves the complicated computation for 3D volume data.The existing hybrid CNN‐Transformer network can only provide the limited performance improvement or even poorer segmentation performance than the pure CNN.To address these issues,a short‐term and long‐term memory self‐attention network is proposed.Firstly,a distinctive self‐attention block uses the Transformer to explore the correlation among the region features at different levels extracted by the CNN.Then,the memory structure filters and combines the above information to exclude the similar regions and detect the multiple tumours.Finally,the multi‐layer reconstruction blocks will predict the tumour boundaries.Experimental results demonstrate that our method outperforms other methods in terms of subjective visual and quantitative evaluation.Compared with the most competitive method,the proposed method provides Dice(82.4%vs.76.6%)and Hausdorff distance 95%(HD95)(10.66 vs.11.54 mm)on the KiTS19 as well as Dice(80.2%vs.78.4%)and HD95(9.632 vs.12.17 mm)on the LiTS.

CoLM^(2)S:Contrastive self‐supervised learning on attributed multiplex graph network with multi‐scale information

Contrastive self‐supervised representation learning on attributed graph networks with Graph Neural Networks has attracted considerable research interest recently.However,there are still two challenges.First,most of the real‐word system are multiple relations,where entities are linked by different types of relations,and each relation is a view of the graph network.Second,the rich multi‐scale information(structure‐level and feature‐level)of the graph network can be seen as self‐supervised signals,which are not fully exploited.A novel contrastive self‐supervised representation learning framework on attributed multiplex graph networks with multi‐scale(named CoLM^(2)S)information is presented in this study.It mainly contains two components:intra‐relation contrast learning and interrelation contrastive learning.Specifically,the contrastive self‐supervised representation learning framework on attributed single‐layer graph networks with multi‐scale information(CoLMS)framework with the graph convolutional network as encoder to capture the intra‐relation information with multi‐scale structure‐level and feature‐level selfsupervised signals is introduced first.The structure‐level information includes the edge structure and sub‐graph structure,and the feature‐level information represents the output of different graph convolutional layer.Second,according to the consensus assumption among inter‐relations,the CoLM^(2)S framework is proposed to jointly learn various graph relations in attributed multiplex graph network to achieve global consensus node embedding.The proposed method can fully distil the graph information.Extensive experiments on unsupervised node clustering and graph visualisation tasks demonstrate the effectiveness of our methods,and it outperforms existing competitive baselines.

Self-Healable and Stretchable PAAc/XG/Bi_(2)Se_(0.3)Te_(2.7) Hybrid Hydrogel Thermoelectric Materials

Thermoelectric power generators have attracted increasing interest in recent years owing to their great potential in wearable electronics power supply.It is noted that thermoelectric power generators are easy to damage in the dynamic service process,resulting in the formation of microcracks and performance degradation.Herein,we prepare a new hybrid hydrogel thermoelectric material PAAc/XG/Bi_(2)Se_(0.3)Te_(2.7)by an in situ polymerization method,which shows a high stretchable and self-healable performance,as well as a good thermoelectric performance.For the sample with Bi_(2)Se_(0.3)Te_(2.7)content of 1.5 wt%(i.e.,PAAc/XG/Bi2Se0.3Te27(1.5 wt%)),which has a room temperature Seebeck coefficient of-0.45 mV K^(-1),and exhibits an open-circuit voltage of-17.91 mV and output power of 38.1 nW at a temperature difference of 40 K.After being completely cut off,the hybrid thermoelectric hydrogel automatically recovers its electrical characteristics within a response time of 2.0 s,and the healed hydrogel remains more than 99%of its initial power output.Such stretchable and self-healable hybrid hydrogel thermoelectric materials show promising potential for application in dynamic service conditions,such as wearable electronics.

Safety and efficacy of Kaffes intraductal self-expanding metal stents in the management of post-liver transplant anastomotic strictures

BACKGROUND Endoscopic management is the first-line therapy for post-liver-transplant anas-tomotic strictures.Although the optimal duration of treatment with plastic stents has been reported to be 8-12 months,data on safety and duration for metal stents in this setting is scarce.Due to limited access to endoscopic retrograde cholan-giopancreatography(ERCP)during the coronavirus disease 2019 pandemic in our centre,there was a change in practice towards increased usage and length-of-stay of the Kaffes biliary intraductal self-expanding stent in patients with suitable anatomy.This was mainly due to the theoretical benefit of Kaffes stents allowing for longer indwelling periods compared to the traditional plastic stents.METHODS Adult liver transplant recipients aged 18 years and above who underwent ERCP were retrospectively identified during a 10-year period through a database query.Unplanned admissions post-Kaffes stent insertion were identified manually through electronic and scanned medical records.The main outcome was the incidence of complications when stents were left indwelling for 3 months vs 6 months.Stent efficacy was calculated via rates of stricture recurrence between patients that had stenting courses for≤120 d or>120 d.RESULTS During the study period,a total of 66 ERCPs with Kaffes insertion were performed in 54 patients throughout their stenting course.In 33 ERCPs,the stent was removed or exchanged on a 3-month interval.No pancreatitis,perfor-ations or deaths occurred.Minor post-ERCP complications were similar between the 3-month(abdominal pain and intraductal migration)and 6-month(abdominal pain,septic shower and embedded stent)groups-6.1%vs 9.1%respectively,P=0.40.All strictures resolved at the end of the stenting course,but the stenting course was variable from 3 to 22 months.The recurrence rate for stenting courses lasting for up to 120 d was 71.4%and 21.4%for stenting courses of 121 d or over(P=0.03).There were 28 patients that were treated with a single ERCP with Kaffes,21 with removal after 120 d and 7 within 120 d.There was a significant improvement in stricture recurrence when the Kaffes was removed after 120 d when a single ERCP was used for the entire stenting course(71.0%vs 10.0%,P=0.01).CONCLUSION Utilising a single Kaffes intraductal fully-covered metal stent for at least 4 months is safe and efficacious for the management of post-transplant anastomotic strictures.

Unified Description of the Three Stable Particles in Self-Action Allows Determination of Their Relative Masses

The Dirac equation γμ(δμ-eAμ)Ψ=mc2Ψ describes the bound states of the electron under the action of external potentials, Aμ. We assumed that the fundamental form of the Dirac equation γμ(δμ-Sμ)Ψ=0 should describe the stable particles (the electron, the proton and the dark-matter-particle (dmp)) bound to themselves under the action of their own potentials Sμ. The new equation reveals that self energy is consequence of self action, it also reveals that the spin angular momentum is consequence of the dynamic structure of the stable particles. The quantitative results are the determination of their relative masses as well as the determination of the electromagnetic coupling constant.

基于轻量化U^(2)-Net的车道线检测算法研究

针对车道线遮挡、道路阴影等多车道驾驶环境下提取的车道线特征信息缺失造成预测车道线模糊、不连续等问题,提出一种基于轻量化U^(2)-Net的车道线检测算法。首先,以轻量化U^(2)-Net的残差U形模块(RSU)和多特征尺度融合获得全局信息丰富的车道线特征;其次,对车道线特征进行逐像素阈值判断,并选择最小二乘法结合感兴趣区域(ROI)中车道线簇进行车道线的拟合,实现多车道线检测并确认自车道线区域;最后,在图森(TuSimple)数据集上进行验证与分析。验证结果表明,所提出的车道线检测算法的平均准确率达到98.4%,相比于其他车道线检测网络,该算法的网络参数量较少,准确率较高。

APPLICATION OF FUZZY LOGIC AND SELF-ORGANIZING NETWORK TO TOOL-WEAR CLASSIFICATION

A tool-wear monitoring system for metal turning operations is presented based on the combinative application of fuzzy logic and unsupervised neural network. A group of self-organizing map （SOM） neural networks is established based on the typical cutting condition combinations, and each of networks is corresponding to a typical cutting condition. For a specifie cutting condition, the fuzzy logic method is used to select an optimum trained SOM network. The proposed monitoring system, ealled the Fuzzy-SOM-TWC, is used to classify tool states based on the in-time measurement of force, aeoustic emission（AE）, and motor eurrent signals. An approximate 98%--100% correct classification of tool-wear status is obtained by testing the system with a series data samples under freely selected cutting conditions.

联合Self-attention与Axial-attention的机场跑道裂缝分割

机场跑道裂缝形态多样、方向各异、长短不一且粗细不均,通常不具有统计规律。现有的各类裂缝分割算法难以在此类复杂场景中落地。针对上述问题,提出了联合self-attention与axial-attention的机场跑道裂缝分割网络(CSA-net),通过引入自注意力模块、轴向注意力模块、可变形卷积模块,提取裂缝的局部特征和全局语义特征。通过transformer decoder还原特征图的原始尺寸,融合了不同尺度间的分割结果,保留尽可能多的细节信息,使得CSA-net有更好的分割精度。在机场跑道实拍的数据集上进行的测试表明,针对裂缝的像素级分割指标F1-score达到了78.91%,高于目前各类裂缝分割算法。

Multipath Source Self Repair Routing Algorithm for Mobile Ad Hoc Networks

A multipath source self repair routing (MSSRR) algorithm for mobile ad hoc networks is proposed. By using multiple paths which can be repaired by themselves to transmit packets alternately, the network's load is balanced, the link state in the network can be checked in time, the number of the times the route discovery mechanism starts is decreased. If only one route which will be broken can be used to transmit the packets, the route discovery mechanism is restarted.The algorithm is implemented on the basis of dynamic source routing (DSR). The effect of MSSRR on lifetime of the access from the source to the destination and the overhead is discussed. Compared with the performance of DSR,it can be seen that the algorithm can improve the performance of the network obviously and the overhead almost does not increase if the average hop count is larger.

Backstepping sliding mode control with self recurrent wavelet neural network observer for a novel coaxial twelve-rotor UAV

The robust attitude control for a novel coaxial twelve-rotor UAV which has much greater payload capacity,higher drive capability and damage tolerance than a quad-rotor UAV is studied. Firstly,a dynamical and kinematical model for the coaxial twelve-rotor UAV is designed. Considering model uncertainties and external disturbances,a robust backstepping sliding mode control( BSMC) with self recurrent wavelet neural network( SRWNN) method is proposed as the attitude controller for the coaxial twelve-rotor. A combinative algorithm of backstepping control and sliding mode control has simplified design procedures with much stronger robustness benefiting from advantages of both controllers. SRWNN as the uncertainty observer is able to estimate the lumped uncertainties effectively.Then the uniformly ultimate stability of the twelve-rotor system is proved by Lyapunov stability theorem. Finally,the validity of the proposed robust control method adopted in the twelve-rotor UAV under model uncertainties and external disturbances are demonstrated via numerical simulations and twelve-rotor prototype experiments.

A Secure-Efficient Data Collection Algorithm Based on Self-Adaptive Sensing Model in Mobile Internet of Vehicles

Existing research on data collection using wireless mobile vehicle network emphasizes the reliable delivery of information.However,other performance requirements such as life cycle of nodes,stability and security are not set as primary design objectives.This makes data collection ability of vehicular nodes in real application environment inferior.By considering the features of nodes in wireless IoV,such as large scales of deployment,volatility and low time delay,an efficient data collection algorithm is proposed for mobile vehicle network environment.An adaptive sensing model is designed to establish vehicular data collection protocol.The protocol adopts group management in model communication.The vehicular sensing node in group can adjust network sensing chain according to sensing distance threshold with surrounding nodes.It will dynamically choose a combination of network sensing chains on basis of remaining energy and location characteristics of surrounding nodes.In addition,secure data collection between sensing nodes is undertaken as well.The simulation and experiments show that the vehicular node can realize secure and real-time data collection.Moreover,the proposed algorithm is superior in vehicular network life cycle,power consumption and reliability of data collection by comparing to other algorithms.

Self-correcting wavelet neural network control of continuous rotary electro-hydraulic servo motor

In allusion to the problem of friction,leakage,vibration and noise existing in continuous rotary motor electro-hydraulic servo system,highly nonlinearity and uncertainties affecting the system performance,based on the transfer function of electro-hydraulic servo system,a kind of Pol-Ind friction model is proposed.The parameters of Pol-Ind friction model are identified and the accurate mathematical model of friction torque is obtained by experiment.The self-correcting wavelet neural network(WNN)controller is proposed,and Adam optimization algorithm is used to perform gradient optimization on scale factor and displacement factor in wavelet basis function,so as to improve the speed and precision of parameter optimization.Through comparative simulation analysis,it is clearly that the self-correcting WNN controller can effectively improve the frequency response and tracking accuracy of continuous rotary motor electro-hydraulic servo system.

Hydraulic Self Servo Swing Cylinder Structure Optimization and Dynamic Characteristics Analysis Based on Genetic Algorithm

The dynamic characteristics of hydraulic self servo swing cylinder were analyzed according to the hydraulic system natural frequency formula. Based on that,a method of the hydraulic self servo swing cylinder structure optimization based on genetic algorithm was proposed in this paper. By analyzing the four parameters that affect the dynamic characteristics, we had to optimize the structure to obtain as larger the Dm( displacement) as possible under the condition with the purpose of improving the dynamic characteristics of hydraulic self servo swing cylinder. So three state equations were established in this paper. The paper analyzed the effect of the four parameters in hydraulic self servo swing cylinder natural frequency equation and used the genetic algorithm to obtain the optimal solution of structure parameters. The model was simulated by substituting the parameters and initial value to the simulink model. Simulation results show that: using self servo hydraulic swing cylinder natural frequency equation to study its dynamic response characteristics is very effective.Compared with no optimization,the overall system dynamic response speed is significantly improved.

Real-valued multi-area self set optimization in immunity-based network intrusion detection system

The real-valued self set in immunity-based network intrusion detection system （INIDS） has some defects： multi-area and overlapping, which are ignored before. The detectors generated by this kind of self set may have the problem of boundary holes between self and nonself regions, and the generation efficiency is low, so that, the self set needs to be optimized before generation stage. This paper proposes a self set optimization algorithm which uses the modified clustering algorithm and Gaussian distribution theory. The clustering deals with multi-area and the Gaussian distribution deals with the overlapping. The algorithm was tested by Iris data and real network data, and the results show that the optimized self set can solve the problem of boundary holes, increase the efficiency of detector generation effectively, and improve the system＇s detection rate.

Handily etching nickel foams into catalyst-substrate fusion self‐stabilized electrodes toward industrial‐level water electrolysis

The key challenge of industrial water electrolysis is to design catalytic electrodes that can stabilize high current density with low power consumption(i.e.,overpotential),while industrial harsh conditions make the balance between electrode activity and stability more difficult.Here,we develop an efficient and durable electrode for water oxidation reaction(WOR),which yields a high current density of 1000 mA cm−2 at an overpotential of only 284 mV in 1M KOH at 25°C and shows robust stability even in 6M KOH strong alkali with an elevated temperature up to 80°C.This electrode is fabricated from a cheap nickel foam(NF)substrate through a simple one-step solution etching method,resulting in the growth of ultrafine phosphorus doped nickel-iron(oxy)hydroxide[P-(Ni,Fe)O_(x)H_(y)]nanoparticles embedded into abundant micropores on the surface,featured as a self-stabilized catalyst–substrate fusion electrode.Such self-stabilizing effect fastens highly active P-(Ni,Fe)O_(x)H_(y)species on conductive NF substrates with significant contribution to catalyst fixation and charge transfer,realizing a win–win tactics for WOR activity and durability at high current densities in harsh environments.This work affords a cost-effective WOR electrode that can well work at large current densities,suggestive of the rational design of catalyst electrodes toward industrial-scale water electrolysis.

基于多层次自注意力机制的U-Net图像分割算法

针对U-Net图像分割在下采样过程中会丢失过多信息且在上采样过程恢复效果不佳,从而导致图像分割精度降低的缺陷,提出了一种基于多层次自注意力机制的U-Net图像分割算法。该多层次自注意力机制在每一层上采样层前均嵌入自注意力模块,将上采样层的输入与缩放的原图拼接后处理成模板图,再与原本的输入信息融合后输出到上采样层。该算法不仅能通过拼接原图的自注意力模块进一步提供更多细节信息,还能利用上采样层的特征选择功能减少拼接原图带来的背景噪音,提高模型的分割精度。最后,在PASCAL VOC数据集和DeepFashion2数据集的基础上进行了人体分割和服装分割实验。实验结果 证明,该方法 能较好地改善图像的分割性能,从而证明了其正确性和有效性。

Parameter Self - Learning of Generalized Predictive Control Using BP Neural Network

This paper describes the self—adjustment of some tuning-knobs of the generalized predictive controller(GPC).A three feedforward neural network was utilized to on line learn two key tuning-knobs of GPC,and BP algorithm was used for the training of the linking-weights of the neural network.Hence it gets rid of the difficulty of choosing these tuning-knobs manually and provides easier condition for the wide applications of GPC on industrial plants.Simulation results illustrated the effectiveness of the method.