Deep Learning-Based Stacked Auto-Encoder with Dynamic Differential Annealed Optimization for Skin Lesion Diagnosis

Intelligent diagnosis approaches with shallow architectural models play an essential role in healthcare.Deep Learning(DL)models with unsupervised learning concepts have been proposed because high-quality feature extraction and adequate labelled details significantly influence shallow models.On the other hand,skin lesionbased segregation and disintegration procedures play an essential role in earlier skin cancer detection.However,artefacts,an unclear boundary,poor contrast,and different lesion sizes make detection difficult.To address the issues in skin lesion diagnosis,this study creates the UDLS-DDOA model,an intelligent Unsupervised Deep Learning-based Stacked Auto-encoder(UDLS)optimized by Dynamic Differential Annealed Optimization(DDOA).Pre-processing,segregation,feature removal or separation,and disintegration are part of the proposed skin lesion diagnosis model.Pre-processing of skin lesion images occurs at the initial level for noise removal in the image using the Top hat filter and painting methodology.Following that,a Fuzzy C-Means(FCM)segregation procedure is performed using a Quasi-Oppositional Elephant Herd Optimization(QOEHO)algorithm.Besides,a novel feature extraction technique using the UDLS technique is applied where the parameter tuning takes place using DDOA.In the end,the disintegration procedure would be accomplished using a SoftMax(SM)classifier.The UDLS-DDOA model is tested against the International Skin Imaging Collaboration(ISIC)dataset,and the experimental results are examined using various computational attributes.The simulation results demonstrated that the UDLS-DDOA model outperformed the compared methods significantly.

Predicting the Antigenic Variant of Human Influenza A(H3N2) Virus with a Stacked Auto-Encoder Model

The influenza virus changes its antigenicity frequently due to rapid mutations, leading to immune escape and failure of vaccination. Rapid determination of the influenza antigenicity could help identify the antigenic variants in time. Here, we built a stacked auto-encoder (SAE) model for predicting the antigenic variant of human influenza A(H3N2) viruses based on the hemagglutinin (HA) protein sequences. The model achieved an accuracy of 0.95 in five-fold cross-validations, better than the logistic regression model did. Further analysis of the model shows that most of the active nodes in the hidden layer reflected the combined contribution of multiple residues to antigenic variation. Besides, some features (residues on HA protein) in the input layer were observed to take part in multiple active nodes, such as residue 189, 145 and 156, which were also reported to mostly determine the antigenic variation of influenza A(H3N2) viruses. Overall,this work is not only useful for rapidly identifying antigenic variants in influenza prevention, but also an interesting attempt in inferring the mechanisms of biological process through analysis of SAE model, which may give some insights into interpretation of the deep learning

A comparative evaluation of Stacked Auto-Encoder neural network and Multi-Layer Extreme Learning Machine for detection and classification of faults in transmission lines using WAMS data

Smart grid is envisaged as a power grid that is extremely reliable and flexible.The electrical grid has wide-area measuring devices like Phasor measurement units(PMUs)deployed to provide real-time grid information and resolve issues effectively and speedily without compromising system availability.The development and application of machine learning approaches for power system protection and state estimation have been facilitated by the availability of measurement data.This research proposes a transmission line fault detection and classification(FD&C)system based on an auto-encoder neural network.A comparison between a Multi-Layer Extreme Learning Machine(ML-ELM)network model and a Stacked Auto-Encoder neural network(SAE)is made.Additionally,the performance of the models developed is compared to that of state-of-the-art classifier models employing feature datasets acquired by wavelet transform based feature extraction as well as other deep learning models.With substantially shorter testing time,the suggested auto-encoder models detect faults with 100% accuracy and classify faults with 99.92% and 99.79%accuracy.The computational efficiency of the ML-ELM model is demonstrated with high accuracy of classification with training time and testing time less than 50 ms.To emulate real system scenarios the models are developed with datasets with noise with signal-to-noise-ratio(SNR)ranging from 10 dB to 40 dB.The efficacy of the models is demonstrated with data from the IEEE 39 bus test system.

Interdigital MnO_(2)/PEDOT Alternating Stacked Microelectrodes for High-Performance On-Chip Microsupercapacitor and Humidity Sensing

For microelectronic devices,the on-chip microsupercapacitors with facile construction and high performance,are attracting researchers'prior consideration due to their high compatibility with modern microsystems.Herein,we proposed interchanging interdigital Au-/MnO_(2)/polyethylene dioxythiophene stacked microsupercapacitor based on a microfabrication process followed by successive electrochemical deposition.The stacked configuration of two pseudocapacitive active microelectrodes meritoriously leads to an enhanced contact area between MnO_(2)and the conductive and electroactive layer of polyethylene dioxythiophene,hence providing excellent electron transport and diffusion pathways of electrolyte ions,resulting in increased pseudocapacitance of MnO_(2)and polyethylene dioxythiophene.The stacked quasi-solid-state microsupercapacitors delivered the maximum specific capacitance of 43 mF cm^(-2)(211.9 F cm^(-3)),an energy density of 3.8μWh cm^(-2)(at a voltage window of 0.8 V)and 5.1μWh cm^(-2)(at a voltage window of 1.0 V)with excellent rate capability(96.6%at 2 mA cm^(-2))and cycling performance of 85.3%retention of initial capacitance after 10000 consecutive cycles at a current density of 5 mA cm^(-2),higher than those of ever reported polyethylene dioxythiophene and MnO_(2)-based planar microsupercapacitors.Benefiting from the favorable morphology,bilayer microsupercapacitor is utilized as a flexible humidity sensor with a response/relaxation time superior to those of some commercially available integrated microsensors.This strategy will be of significance in developing high-performance on-chip integrated microsupercapacitors/microsensors at low cost and environment-friendly routes.

SSA-over-array(SSoA):A stacked DRAM architecture for nearmemory computing

Aiming to enhance the bandwidth in near-memory computing,this paper proposes a SSA-over-array(SSoA)architecture.By relocating the secondary sense amplifier(SSA)from dynamic random access memory(DRAM)to the logic die and repositioning the DRAM-to-logic stacking interface closer to the DRAM core,the SSoA overcomes the layout and area limitations of SSA and master DQ(MDQ),leading to improvements in DRAM data-width density and frequency,significantly enhancing bandwidth density.The quantitative evaluation results show a 70.18 times improvement in bandwidth per unit area over the baseline,with a maximum bandwidth of 168.296 Tbps/Gb.We believe the SSoA is poised to redefine near-memory computing development strategies.

Fault Diagnosis of Motor in Frequency Domain Signal by Stacked De-noising Auto-encoder

With the rapid development of mechanical equipment,mechanical health monitoring field has entered the era of big data.Deep learning has made a great achievement in the processing of large data of image and speech due to the powerful modeling capabilities,this also brings influence to the mechanical fault diagnosis field.Therefore,according to the characteristics of motor vibration signals(nonstationary and difficult to deal with)and mechanical‘big data’,combined with deep learning,a motor fault diagnosis method based on stacked de-noising auto-encoder is proposed.The frequency domain signals obtained by the Fourier transform are used as input to the network.This method can extract features adaptively and unsupervised,and get rid of the dependence of traditional machine learning methods on human extraction features.A supervised fine tuning of the model is then carried out by backpropagation.The Asynchronous motor in Drivetrain Dynamics Simulator system was taken as the research object,the effectiveness of the proposed method was verified by a large number of data,and research on visualization of network output,the results shown that the SDAE method is more efficient and more intelligent.

ConvNeXt网络及Stacked BiLSTM-Self-Attention在轴承剩余寿命预测中的应用

在滚动轴承剩余使用寿命预测方面,采用传统方法时存在鲁棒性差、精度低等各种问题。近些年来深度学习的发展为解决这些问题提供了新的思路。为了进一步提高对轴承寿命的预测精度,提出了一种基于ConvNeXt网络、堆叠双向长短时记忆网络(SBiLSTM)和自注意力机制(Self-Attention)的滚动轴承寿命预测方法。首先,采用连续小波变换(CWT)构造了振动信号的时频图,以更好地捕捉信号的时域和频域特征;然后,将得到的时频图输入到构建的ConvNeXt网络中,通过卷积、池化和层归一化等操作,对时频图的关键特征进行了提取;最后,将提取后的特征输入到SBiLSTM-Self-Attention模块中,进一步提取了时序信息和特征权重分配数据,利用PHM2012挑战数据集进行了验证,通过实验分析了该方法的均方根误差(RMSE)和平均绝对误差(MAE)。研究结果表明:相较于现有技术方法,该方法的平均RMSE为0.031;与其他三种方法,即卷积神经网络(CNN)、深度残差双向门控循环单元(DRN-BiGRU)和深度卷积自注意力双向门控循环单元(DCNN-Self-Attention-BiGRU)相比,其平均RMSE值分别下降了79%、74%和55%,MAE值分别下降了78%、73%和53%,说明该方法在滚动轴承剩余寿命预测中有较好的性能。

Fault Diagnosis for Rolling Bearings with Stacked Denoising Auto-encoder of Information Aggregation

Rolling bearings are important central components in rotating machines, whose fault diagnosis is crucial in condition-based maintenance to reduce the complexity of different kinds of faults. To classify various rolling bearing faults, a prognostic algorithm consisting of four phases was proposed. Since stacked denoising auto-encoder can be filtered, noise of large numbers of mechanical vibration signals was used for deep learning structure to extract the characteristics of the noise. Unsupervised pre-training method, which can greatly simplify the traditional manual extraction approach, was utilized to process the depth of the data automatically. Furthermore, the aggregation layer of stacked denoising auto-encoder(SDA) was proposed to get rid of gradient disappearance in deeper layers of network, mix superficial nodes’ expression with deeper layers, and avoid the insufficient express ability in deeper layers. Principal component analysis(PCA) was adopted to extract different features for classification. According to the experimental data of this method and from the comparison results, the proposed method of rolling bearing fault classification reached 97.02% of correct rate, suggesting a better performance than other algorithms.

基于动态Stacked-GBDT算法的数据资源价值评估方法研究

针对现有的数据资源价值评估与定价方法主观性强、定量标准缺乏的问题,提出基于模型堆叠集成GBDT(Stacked-GBDT)算法的数据资源价值评估方法。首先,基于敏感性分析,从数据自身和市场两个维度归纳并建立了数据资源价值评估指标体系;然后,基于GBDT机器学习算法与Stacking集成学习算法,提出了基于StackedGBDT的数据资源价值评估算法,并与Random Forest和XGBoost算法进行对比以验证所提方法的正确性及有效性;最后,应用Stacked-GBDT模型对数据集进行动态定价。结果表明,Stacked-GBDT算法构建的数据资源价值评估模型可为数据价值测算及动态定价提供精确可靠的依据与支撑。

基于Stacking融合模型的Web攻击检测方法

随着计算机技术与互联网技术的飞速发展,Web应用在人们的生产与生活中扮演着越来越重要的角色。但是在人们的日常生活与工作中带来了更多便捷的同时,却也带来了严重的安全隐患。在开发Web应用的过程中,大量不规范的新技术应用引入了很多的网站漏洞。攻击者可以利用Web应用开发过程中的漏洞发起攻击,当Web应用受到攻击时会造成严重的数据泄露和财产损失等安全问题,因此Web安全问题一直受到学术界和工业界的关注。超文本传输协议(HTTP)是一种在Web应用中广泛使用的应用层协议。随着HTTP协议的大量使用,在HTTP请求数据中包含了大量的实际入侵,针对HTTP请求数据进行Web攻击检测的研究也开始逐渐被研究人员所重视。本文提出了一种基于Stacking融合模型的Web攻击检测方法,针对每一条文本格式的HTTP请求数据,首先进行格式化处理得到既定的格式,结合使用Word2Vec方法和TextCNN模型将其转换成向量化表示形式;然后利用Stacking模型融合方法,将不同的子模型(使用配置不同尺寸过滤器的Text-CNN模型搭配不同的检测算法)进行融合搭建出Web攻击检测模型,与融合之前单独的子模型相比在准确率、召回率、F1值上都有所提升。本文所提出的Web攻击检测模型在公开数据集和真实环境数据上都取得了更加稳定的检测性能。

基于Stacking融合模型的PHEV复合储能系统实时能量分配策略

为了解决插电式混合动力汽车单一电池低比功率、无法响应暂态功率需求的问题,设计了电池和超级电容并联式复合储能系统。同时针对采用动态规划法优化负载电流分配时缺乏实时性的问题,利用不同驱动工况下动态规划优化的结果构成训练集进行训练,并综合GRU网络以及XGBoost算法,提出了一种Stacking集成学习框架下多模型融合的能量分配策略。仿真结果表明,与仅使用单一电池的储能系统相比,基于Stacking融合模型的实时能量分配系统在UDDS和US06两种循环工况下,电池峰值电流分别降低了48.7%和50.8%,有效削弱了电池的峰值电流,提升了电池的整体性能。

基于Stacking集成学习的热轧带钢凸度诊断模型

在热连轧生产过程中,凸度是重要的质量指标,过程数据的非平衡性限制了数据驱动模型的预测效果,为提高模型的预测精度,提出一种融合SMOTE和Stacking集成算法的热轧带钢凸度诊断模型。首先,采用SMOTE过采样方法处理凸度相关数据集,降低数据非平衡分布导致的影响;然后,构建以轻量级梯度提升机(LightGBM)、支持向量机(SVM)、K近邻(KNN)和随机森林(RF)为基学习器,逻辑回归(LR)为元学习器的Stacking集成模型,最后,使用某2160 mm热轧带钢实际生产数据进行模型验证。研究结果表明,诊断模型的准确率、少数类召回率、平衡F分数、几何平均值和ROC曲线下面积分别为0.9580、0.9595、0.9573、0.9589和0.9579,与XGBoost、LightGBM、KNN、SVM和随机森林模型对比,预测效果最优,证明了Stacking集成算法能够有效增强诊断模型的泛化能力,具有优良的诊断性能。

近红外光谱结合Stacking集成学习的猕猴桃糖度检测研究

利用近红外光谱技术Stacking集成学习对猕猴桃糖度的无损检测。以湖北“云海一号”猕猴桃为研究对象,采用红外分析仪获取了280个样本的光谱数据,包含了4000~10000cm^(-1)范围内的1557个波长数据,使用折射仪测量糖度值。通过蒙特卡洛随机采样结合T检验的奇异样本识别算法筛除异常值样本。利用SPXY算法按照4∶1的比例划分训练集和测试集。使用多元散射校正(MSC)、SG平滑滤波(SG)、趋势校正(DT)、矢量归一化(VN)、标准正态变换(SNV)五种方法对数据进行预处理。使用无信息变量消除法(UVE)、竞争性自适应重加权算法(CARS)和区间变量迭代空间收缩特征选择算法(iVISSA)提取特征波长,使用连续投影算法(SPA)进行二次提取,消除共线性变量。由于单一模型的泛化能力有限,为了扩大建模能力,设计了一种基于Stacking算法的集成学习模型。选择贝叶斯岭回归(BRR)、偏最小二乘回归(PLSR)、支持向量机回归(SVR)以及人工神经网络(ANN)作为基学习器,线性回归(LR)作为元学习器建立集成模型,比较不同组合下集成模型的性能。使用Pearson相关系数分析基学习器与集成模型之间的关系。结果表明:在五种预处理方法之中,矢量归一化的效果最佳。对预处理后的光谱进行特征波长提取,结果显示VN-CARS-PLSR模型效果最好,在测试集上的RP2为0.805,RMSEP为0.498。模型提取了177个特征波长,数据量相比于原始光谱减少了88.6%。通过Stacking算法对基学习器进行融合,对比不同的组合方式,发现PLS+SVR+ANN集成模型预测精度最高,RP2达到了0.853,RMSEP下降至0.433。通过Pearson相关系数分析了基学习器对集成模型性能的影响。研究表明,与单一模型相比,Stacking集成模型能够进行更加全面的建模,具有更高的泛化能力,该方法为猕猴桃糖度品质的无损检测提供了技术支持。

基于组合时域特征提取和Stacking集成学习的燃煤锅炉NO_(x)排放浓度预测

为提高火电厂锅炉出口NO_(x)排放浓度的预测精度,提出一种考虑组合时域特征的Stacking集成学习模型。首先,为挖掘数据深层信息,采用时序分析、完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise analysis,CEEMDAN)和统计学计算数据标准差、偏度等特征的方法进行组合时域特征提取以构建重构数据;其次,考虑到重构数据中存在的冗余变量对模型的精度有所影响,利用遗传算法(genetic algorithm,GA)对重构数据进行特征降维;最后,为充分发挥各个模型的优势以提高模型的预测精度,构建以极限学习机(extreme learning machines,ELM)、深度神经网络(deep neural networks,DNN)、多层感知器(multilayer perceptron,MLP)、极限梯度提升算法(extreme gradient boosting,XGBoost)为基模型和以回声状态网络(echo state network,ESN)为元模型的Stacking集成学习NOx排放浓度预测模型。实验结果表明:该预测模型在不同数据集下都有着不错的预测效果,预测误差均小于2%,能够对锅炉NOx排放浓度实现精准预测。

基于FIR-Stacking的刀具磨损预测

针对铣刀加工工件时传感器信号存在噪声、单一传统机器学习模型预测效果不理想的问题,提出一种基于自适应FIR滤波器和Stacking集成模型的刀具磨损预测方法。首先,采用自适应FIR滤波器去噪,计算时域、频域和时频域常用统计量作为信号特征,并对同一信号的多源信号特征进行拼接,经Pearson相关系数筛选保留相关系数大于0.2的特征;最后,以LightGBM、支持向量回归(support vector regression,SVR)、多层感知机(multilayer perceptron,MLP)作为基模型,Lasso作为元模型,构建Stacking集成模型进行刀具磨损预测。使用铣削加工数据集进行验证,结果表明该方法可有效提高预测准确性。

基于Stacking算法集成学习的页岩油储层总有机碳含量评价方法

总有机碳含量(TOC)是页岩油储层评价的重要参数,而传统总有机碳含量测井评价方法精度较低且普适性较差,机器学习模型在一定程度上提高了总有机碳含量预测精度,但结果存在不稳定性。为了进一步提高页岩油储层总有机碳含量预测精度,基于有机质岩石物理特征和不同总有机碳含量测井响应特征的深入分析,优选出深侧向电阻率、声波时差、补偿中子和密度测井曲线作为总有机碳含量的敏感测井响应,并将其作为输入特征,以岩心分析总有机碳含量作为期望输出值,分别建立了决策树模型、支持向量回归机模型、BP(Back Propagation)神经网络模型,并建立了以决策树模型为基模型、支持向量回归机模型为元模型的Stacking算法集成学习模型。利用B油田A区块的岩心样本数据和实际井数据对不同模型预测总有机碳含量结果进行了验证,结果表明,基于Stacking算法的集成学习模型的总有机碳含量预测精度最高,相较于决策树模型、支持向量回归机模型、BP神经网络模型和改进的ΔlgR法,预测精度有较大提高。因此,基于Stacking算法的集成学习模型为该研究区最有效的总有机碳含量计算方法,这为准确地评估页岩油储层的生烃潜力、确保页岩油储层的高效开采及资源利用奠定了基础。

基于Stacking集成学习的声波时差测井曲线复原研究

声波时差测井曲线在石油勘探中发挥着不可或缺的作用,但是受地质或仪器的影响,经常会出现部分甚至完整的声波测井曲线缺失的情况。针对这一问题,提出了一种基于Stacking集成学习的声波时差测井曲线复原方法,该模型使用随机森林(RF)、梯度提升决策树(GBDT)、轻量梯度提升机(LightGBM)和极限梯度提升(XGBoost)作为基学习器,支持向量回归(SVR)作为元学习器,同时采用5折交叉验证的方法。实验选取了大庆油田某区块的实际测井数据,分别进行了同井和异井间的缺失声波时差测井曲线复原实验,结果表明,所提方法比单一模型预测更加准确,验证了此方法的可行性。

基于Stacking集成算法的抛石护岸水毁破坏预测研究

抛石护岸在顶冲等极端情况下易发生水毁破坏,给人民的生命财产带来威胁。通过水槽试验获取496组样本数据,利用互信息(MI)筛选出6个关键特征属性,并采用支持向量机(SVR)、广义回归神经网络(GRNN)和随机森林(RF)等机器学习算法构建多个预测模型。然后,将这些模型作为基学习器,结合BP神经网络(BPNN)作为元学习器,采用Stacking集成学习方法构建抛石护岸破坏程度预测模型。最后,通过决定系数(R^(2))、均方根误差(R_(RMSE))及平均绝对误差(M_(MAE))等评价指标对模型性能进行评估。结果表明,Stacking模型在抛石护岸破坏高度、长度、范围上的平均R^(2)为0.98、RRMSE为0.02、M_(MAE)为0.03,相较于单一模型(SVR、GRNN、RF),Stacking模型的R_(RMSE)、M_(MAE)皆为最小,R2最高。在抛石护岸水毁破坏程度的预测中,融合的Stacking模型展现出更高的准确性与稳定性。

基于Stacking模型的学术论文多标签分类系统构建

学术论文高质量多标签自动分类是推动学术研究发展的关键程序之一。本研究利用Stacking模型将随机森林、支持向量机、极限树、极端梯度提升和神经网络五个分类器融合为一个异质集成分类器,并利用基于问题转换思想的多二分类模型将该分类器应用于学术论文多标签分类。根据学术论文的特点,依次实现了与之配套的论文特征提取模块、TF-IDF加权模块、数据预处理模块,最终构建成一个面向学术论文的多标签分类系统。仿真实验验证了本研究构建的学术论文多标签分类系统在处理学术论文多标签分类问题时,较传统的单模型分类器或同质集成模型分类器在泛化能力、稳定性与准确率方面都有一定程度的提升。图9。参考文献21。

基于互补集成经验模态分解和Stacking融合的短期风速组合预测模型

针对风电场短期风速预测准确度不高的问题,提出一种基于互补集成经验模态分解和Stacking融合的短期风速组合预测模型。首先,为突出短期风速的局部特征并降低建模难度,通过互补集成经验模态分解算法将短期风速分解为若干个稳定分量。然后,利用信息熵和近似熵来判定各分量的复杂度,高复杂度分量选择最小二乘支持向量机、低复杂度分量选择随机配置网络作为对应的预测模型。利用Stacking算法对每个模型的预测值进行融合,使预测精度得到提升。最后,通过一组实际的短期风速数据作为研究对象,将提出的预测模型应用于其预测。对比结果表明,所提预测模型可提高短期风速的预测精度。