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。

Rock mass quality classification based on deep learning:A feasibility study for stacked autoencoders

Objective and accurate evaluation of rock mass quality classification is the prerequisite for reliable sta-bility assessment.To develop a tool that can deliver quick and accurate evaluation of rock mass quality,a deep learning approach is developed,which uses stacked autoencoders(SAEs)with several autoencoders and a softmax net layer.Ten rock parameters of rock mass rating(RMR)system are calibrated in this model.The model is trained using 75%of the total database for training sample data.The SAEs trained model achieves a nearly 100%prediction accuracy.For comparison,other different models are also trained with the same dataset,using artificial neural network(ANN)and radial basis function(RBF).The results show that the SAEs classify all test samples correctly while the rating accuracies of ANN and RBF are 97.5%and 98.7%,repectively,which are calculated from the confusion matrix.Moreover,this model is further employed to predict the slope risk level of an abandoned quarry.The proposed approach using SAEs,or deep learning in general,is more objective and more accurate and requires less human inter-vention.The findings presented here shall shed light for engineers/researchers interested in analyzing rock mass classification criteria or performing field investigation.

Numerical Simulation Investigation on Split Sleeve Cold Expansion of Ti-Al Stacked Structure

Split sleeve cold expansion(SSCX)can efiectively enhance fatigue life of holes by improving the field of residual stress.Numerical simulations were conducted to investigate the parameter influence mechanism and obtain higher compressive residual stress(CRS).Expansion method,degree of cold expansion(DCE),friction coefficient between laminations and depth-diameter ratio were analyzed.For Ti-Al stacked joint holes,two expansion methods are proposed,namely aluminum alloy first followed titanium alloy(Al first)and titanium alloy first followed aluminum alloy(Ti first).The results show that expansion method and DCE have significant efiects on the field of circumferential residual stress,and the friction has a negligible influence.A higher value of CRS and a wider layer of plastic deformation are induced with Ti first.Optimal DCE of TiAl stacked structure is 5.2%-5.6%.As the depth-diameter ratio is in the range of 0.5-1.25,a positive linear correlation between the maximum compressive residual stress(CRS_(max))and depth-diameter ratio is shown.

Layer stacked SiO_(x) microparticle with disconnected interstices enables stable interphase and particle integrity for lithium-ion batteries

Severe mechanical fractu re and unstable interphase,associated with the large volumetric expansion/contraction,significantly hinder the application of high-capacity SiO_(x)materials in lithium-ion batteries.Herein,we report the design and facile synthesis of a layer stacked SiO_(x)microparticle(LS-SiO_(x))material,which presents a stacking structure of SiO_(x)layers with abundant disconnected interstices.This LS-SiO_(x)microparticle can effectively accommodate the volume expansion,while ensuring negligible particle expansion.More importantly,the interstices within SiO_(x)microparticle are disconnected from each other,which efficiently prevent the electrolyte from infiltration into the interior,achieving stable electrode/-electrolyte interface.Accordingly,the LS-SiO_(x)material without any coating delivers ultrahigh average Coulombic efficiency,outstanding cycling stability,and full-cell applicability.Only 6 cycles can attain>99.92%Coulombic efficiency and the capacity retention at 0.05 A g^(-1)for 100 cycles exceeds99%.After 800 cycles at 1 A g^(-1),the thickness swelling of LS-SiO_(x)electrode is as low as 0.87%.Moreover,the full cell with pure LS-SiO_(x)anode exhibits capacity retention of 91.2%after 300 cycles at 0.2 C.This work provides a novel concept and effective approach to rationally design silicon-based and other electrode materials with huge volume variation for electrochemical energy storage applications.

Alternatingly stacked thin film electrodes-based compact aqueous hybrid electrochemical capacitors for hundred-volts AC line filtering

Filtering capacitor with compact configuration and a wide range of operating voltage has been attracting increasing attention for the smooth conversion of the electric signal in modern circuits.Lossless integration of capacitor units can be regarded as one of the efficient ways to achieve a wider voltage range,which has not yet been fully conquered due to the lack of rational designs of the electrode structure and integration technology.This study presents an alternatingly stacked assemble technology to conveniently fabricate compact aqueous hybrid integrated filtering capacitors on a large scale,in which a unit consists of rGO/MXene composite film as a negative electrode and PEDOT:PSS based film as a positive electrode.Benefiting from the synergistic effect of rGO and MXene components,and morphological characteristics of PEDOT:PSS,the capacitor unit exhibits outstanding AC line filtering with a large areal specific energy density of 1,015 μF V^(2)cm^(-2)(0.28 μW h cm^(-2)) at 120 Hz.After rational integration,the assembled capacitors present compact/lightweight configuration and lossless frequency response,as reflected by almost constant resistor-capacitor time constant of 0.2 ms and dissipation factor of 15% at120 Hz,identical to those of the single capacitor unit.Apart from standing alone steadily on a flower,a small volume(only 8.1 cm^(3)) of the integrated capacitor with 70 units connected in series achieves hundred-volts alternating current line filtering,which is superior to most reported filtering capacitors with sandwich configuration.This study provides insight into the fabrication and application of compact/ultralight filtering capacitors with lossless frequency response,and a wide range of operating voltage.

Active straining engineering on self-assembled stacked Ni-based hybrid electrode for ultra-low overpotential

Generating sufficient strains on metal surfaces are highly challenging owing to that most metals can deform plastically to relax the strains on the surfaces.In this work,we developed a facile but highly efficient stacked deposition strategy to in situ activation and reconstruction of NiO/NiOOH on Ni matrix,following with the migration of Fe ions to NiOOH.The Fe sites on the Ni/NiO/NiOOH facilitate the formation of the stable*OH oxygenated intermediates,and the Ni matrix in the catalyst provides the catalyst excellent stability.The oxygen evolution reaction(OER)performance of the stacked NiFe-5 with compressive strain displays the strengthened binding to oxygenated intermediates and superior OER activity,the ultralow overpotentials of 162 versus reversible hydrogen electrode at 10 mA cm^(-2).On the other hand,the Ni-5 without the incorporation of Fe has shown an outstanding hydrogen evolution reaction(HER)activity,affording an overpotential of 47 mV at 10 mA cm^(-2).The NiFe-5‖Ni-5 enables the overall water splitting at a voltage of 1.508 V to achieve 20 mA cm^(-2) with remarkable durability.The stacked deposition strategy improves binding strength of Ni-based catalysts to oxygenated intermediates via generating compressive strain,causing high catalytic activities on OER and HER.

A Stacked Ensemble Deep Learning Approach for Imbalanced Multi-Class Water Quality Index Prediction

A common difficulty in building prediction models with real-world environmental datasets is the skewed distribution of classes.There are significantly more samples for day-to-day classes,while rare events such as polluted classes are uncommon.Consequently,the limited availability of minority outcomes lowers the classifier’s overall reliability.This study assesses the capability of machine learning(ML)algorithms in tackling imbalanced water quality data based on the metrics of precision,recall,and F1 score.It intends to balance the misled accuracy towards the majority of data.Hence,10 ML algorithms of its performance are compared.The classifiers included are AdaBoost,SupportVector Machine,Linear Discriminant Analysis,k-Nearest Neighbors,Naive Bayes,Decision Trees,Random Forest,Extra Trees,Bagging,and the Multilayer Perceptron.This study also uses the Easy Ensemble Classifier,Balanced Bagging,andRUSBoost algorithm to evaluatemulti-class imbalanced learning methods.The comparison results revealed that a highaccuracy machine learning model is not always good in recall and sensitivity.This paper’s stacked ensemble deep learning(SE-DL)generalization model effectively classifies the water quality index(WQI)based on 23 input variables.The proposed algorithm achieved a remarkable average of 95.69%,94.96%,92.92%,and 93.88%for accuracy,precision,recall,and F1 score,respectively.In addition,the proposed model is compared against two state-of-the-art classifiers,the XGBoost(eXtreme Gradient Boosting)and Light Gradient Boosting Machine,where performance metrics of balanced accuracy and g-mean are included.The experimental setup concluded XGBoost with a higher balanced accuracy and G-mean.However,the SE-DL model has a better and more balanced performance in the F1 score.The SE-DL model aligns with the goal of this study to ensure the balance between accuracy and completeness for each water quality class.The proposed algorithm is also capable of higher efficiency at a lower computational time against using the standard SyntheticMinority Oversampling Technique(SMOTE)approach to imbalanced datasets.

Feature Enhanced Stacked Auto Encoder for Diseases Detection in Brain MRI

The detection of brain disease is an essential issue in medical and research areas.Deep learning techniques have shown promising results in detecting and diagnosing brain diseases using magnetic resonance imaging(MRI)images.These techniques involve training neural networks on large datasets of MRI images,allowing the networks to learn patterns and features indicative of different brain diseases.However,several challenges and limitations still need to be addressed further to improve the accuracy and effectiveness of these techniques.This paper implements a Feature Enhanced Stacked Auto Encoder(FESAE)model to detect brain diseases.The standard stack auto encoder’s results are trivial and not robust enough to boost the system’s accuracy.Therefore,the standard Stack Auto Encoder(SAE)is replaced with a Stacked Feature Enhanced Auto Encoder with a feature enhancement function to efficiently and effectively get non-trivial features with less activation energy froman image.The proposed model consists of four stages.First,pre-processing is performed to remove noise,and the greyscale image is converted to Red,Green,and Blue(RGB)to enhance feature details for discriminative feature extraction.Second,feature Extraction is performed to extract significant features for classification using DiscreteWavelet Transform(DWT)and Channelization.Third,classification is performed to classify MRI images into four major classes:Normal,Tumor,Brain Stroke,and Alzheimer’s.Finally,the FESAE model outperforms the state-of-theart,machine learning,and deep learning methods such as Artificial Neural Network(ANN),SAE,Random Forest(RF),and Logistic Regression(LR)by achieving a high accuracy of 98.61% on a dataset of 2000 MRI images.The proposed model has significant potential for assisting radiologists in diagnosing brain diseases more accurately and improving patient outcomes.

Bi-LSTM-Based Deep Stacked Sequence-to-Sequence Autoencoder for Forecasting Solar Irradiation and Wind Speed

Wind and solar energy are two popular forms of renewable energy used in microgrids and facilitating the transition towards net-zero carbon emissions by 2050.However,they are exceedingly unpredictable since they rely highly on weather and atmospheric conditions.In microgrids,smart energy management systems,such as integrated demand response programs,are permanently established on a step-ahead basis,which means that accu-rate forecasting of wind speed and solar irradiance intervals is becoming increasingly crucial to the optimal operation and planning of microgrids.With this in mind,a novel“bidirectional long short-term memory network”(Bi-LSTM)-based,deep stacked,sequence-to-sequence autoencoder(S2SAE)forecasting model for predicting short-term solar irradiation and wind speed was developed and evaluated in MATLAB.To create a deep stacked S2SAE prediction model,a deep Bi-LSTM-based encoder and decoder are stacked on top of one another to reduce the dimension of the input sequence,extract its features,and then reconstruct it to produce the forecasts.Hyperparameters of the proposed deep stacked S2SAE forecasting model were optimized using the Bayesian optimization algorithm.Moreover,the forecasting performance of the proposed Bi-LSTM-based deep stacked S2SAE model was compared to three other deep,and shallow stacked S2SAEs,i.e.,the LSTM-based deep stacked S2SAE model,gated recurrent unit-based deep stacked S2SAE model,and Bi-LSTM-based shallow stacked S2SAE model.All these models were also optimized and modeled in MATLAB.The results simulated based on actual data confirmed that the proposed model outperformed the alternatives by achieving an accuracy of up to 99.7%,which evidenced the high reliability of the proposed forecasting.

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.

New Approach of Multi-Cell Stacked Cell Inverter for Solar Photovoltaic System

In this paper, a new inverter topology dedicated to isolated or grid-connected PV systems is proposed. This inverter is based on the structures of a stacked multi-cell converter (SMC) and an H-bridge. This new topology has allowed the voltage stresses of the converter to be distributed among several switching cells. Secondly, divide the input voltage into several fractions to reduce the number of power semiconductors to be switched. In this contribution, the general topology of this micro-inverter has been described and the simulation tests developed to validate its operation have been presented. Finally, we discussed the simulation results, the efficiency of this topology and the feasibility of its use in a grid-connected photovoltaic production system.

An Efficient Stacked Ensemble Model for Heart Disease Detection and Classification

Cardiac disease is a chronic condition that impairs the heart’s functionality.It includes conditions such as coronary artery disease,heart failure,arrhythmias,and valvular heart disease.These conditions can lead to serious complications and even be life-threatening if not detected and managed in time.Researchers have utilized Machine Learning(ML)and Deep Learning(DL)to identify heart abnormalities swiftly and consistently.Various approaches have been applied to predict and treat heart disease utilizing ML and DL.This paper proposes a Machine and Deep Learning-based Stacked Model(MDLSM)to predict heart disease accurately.ML approaches such as eXtreme Gradient Boosting(XGB),Random Forest(RF),Naive Bayes(NB),Decision Tree(DT),and KNearest Neighbor(KNN),along with two DL models:Deep Neural Network(DNN)and Fine Tuned Deep Neural Network(FT-DNN)are used to detect heart disease.These models rely on electronic medical data that increases the likelihood of correctly identifying and diagnosing heart disease.Well-known evaluation measures(i.e.,accuracy,precision,recall,F1-score,confusion matrix,and area under the Receiver Operating Characteristic(ROC)curve)are employed to check the efficacy of the proposed approach.Results reveal that the MDLSM achieves 94.14%prediction accuracy,which is 8.30%better than the results from the baseline experiments recommending our proposed approach for identifying and diagnosing heart disease.

Saturation thickness of stacked SiO_(2)in atomic-layer-deposited Al_(2)O_(3)gate on 4H-SiC

High-k materials as an alternative dielectric layer for SiC power devices have the potential to reduce interfacial state defects and improve MOS channel conduction capability.Besides,under identical conditions of gate oxide thickness and gate voltage,the high-k dielectric enables a greater charge accumulation in the channel region,resulting in a larger number of free electrons available for conduction.However,the lower energy band gap of high-k materials leads to significant leakage currents at the interface with Si C,which greatly affects device reliability.By inserting a layer of SiO_(2)between the high-k material and Si C,the interfacial barrier can be effectively widened and hence the leakage current will be reduced.In this study,the optimal thickness of the intercalated SiO_(2)was determined by investigating and analyzing the gate dielectric breakdown voltage and interfacial defects of a dielectric stack composed of atomic-layer-deposited Al_(2)O_(3)layer and thermally nitride SiO_(2).Current-voltage and high-frequency capacitance-voltage measurements were performed on metal-oxide-semiconductor test structures with 35 nm thick Al_(2)O_(3)stacked on 1 nm,2 nm,3 nm,6 nm,or 9 nm thick nitride SiO_(2).Measurement results indicated that the current conducted through the oxides was affected by the thickness of the nitride oxide and the applied electric field.Finally,a saturation thickness of stacked SiO_(2)that contributed to dielectric breakdown and interfacial band offsets was identified.The findings in this paper provide a guideline for the SiC gate dielectric stack design with the breakdown strength and the interfacial state defects considered.

基于BERT_Stacked LSTM的农业病虫害问句分类方法

为解决农业病虫害问句分类过程中存在公开数据集较少、文本较短、特征稀疏、隐含语义信息较难学习等问题,以火爆农资招商网为数据源,构建了用于农业病虫害问句分类的数据集,提出了一种用于农业病虫害问句分类的深度学习模型BERT;tacked LSTM。首先,BERT部分获取各个问句的字符级语义信息,生成了包含句子级特征信息的隐藏向量。然后,使用堆叠长短期记忆网络(Stacked LSTM)学习到隐藏的复杂语义信息。实验结果表明,与其他对比模型相比,本文模型对农业病虫害问句分类更具优势,F1值达到了95.76%,并在公开通用领域数据集上进行了测试,F1值达到了98.44%,表明了模型具有较好的的泛化性。

Transgenic restorer rice line T1c-19 with stacked cry1C*/bar genes has low weediness potential without selection pressure

Stacked（insect and herbicide resistant） transgenic rice T1c-19 with cry1C＊/bar genes, its receptor rice Minghui 63（herein MH63） and a local two-line hybrid indica rice Fengliangyou Xiang 1（used as a control） were compared for agronomic performance under field conditions without the relevant selection pressures. Agronomic traits（plant height, tiller number, and aboveground dry biomass）, reproductive ability（pollen viability, panicle length, and filled grain number of main panicles, seed set, and grain yield）, and weediness characteristics（seed shattering, seed overwintering ability, and volunteer seedling recruitment） were used to assess the potential weediness without selection pressure of stacked transgene rice T1c-19. In wet direct-seeded and transplanted rice fields, T1c-19 and its receptor MH63 performed similarly regarding vegetative growth and reproductive ability, but both of them were significantly inferior to the control. T1c-19 did not display weed characteristics; it had weak overwintering ability, low seed shattering and failed to establish volunteers. Exogenous insect and herbicide resistance genes did not confer competitive advantage to transgenic rice T1c-19 grown in the field without the relevant selection pressures.

Fitness of F1 hybrids between stacked transgenic rice T1c-19 with cry1C*/bar genes and weedy rice

Compared to single-trait transgenic crops, stacked transgenic plants may be more prone to become weedy, and transgene flow from stacked transgenic plants to weedy relatives may pose a potential environmental risk because these hybrids could be more advantageous under specific environmental conditions. Evaluation of the potential environmental risk caused by stacked transgenes is essential for assessing the environmental consequences caused by crop-weed transgene flow. The agronomic performance of fitness-related traits was assessed in F1+(transgene positive) hybrids(using the transgenic line T1 c-19 as the paternal parent) in monoculture and mixed planting under presence or absence glufosinate pressure in the presence or absence of natural insect pressure and then compared with the performance of F1–(transgene negative) hybrids(using the non-transgenic line Minghui 63(MH63) as the paternal parent) and their weedy rice counterparts. The results demonstrated that compared with the F1– hybrids and weedy rice counterparts, the F1+ hybrid presented higher performance(P<0.05) or non-significant changes(P>0.05) under natural insect pressure, respectively, lower performance(P<0.05) or non-significant changes(P>0.05) in the absence of insect pressure in monoculture planting, respectively. And compared to weedy rice counterparts, the F1+ hybrid presented higher performance(P<0.05) or non-significant changes(P>0.05) in the presence or absence of insect pressure in mixed planting, respectively. The F1+ hybrids presented nonsignificant changes(P>0.05) under the presence or absence glufosinate pressure under insect or non-insect pressure in monoculture planting. The all F1+ hybrids and two of three F1– hybrids had significantly lower(P<0.05) seed shattering than the weedy rice counterparts. The potential risk of gene flow from T1 c-19 to weedy rice should be prevented due to the greater fitness advantage of F1 hybrids in the majority of cases.

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.

Neutronics analysis of a stacked structure for a subcritical system with LEU solution driven by a D-T neutron source for~(99)Mo production

The utilization of neutrons markedly affects the medical isotope yield of a subcritical system driven by an external D-T neutron source.The general methods to improve the utilization of neutrons include moderating multiplying,and reflecting neutrons,which ignores the use of neutrons that backscatter to the source direction.In this study,a stacked structure was formed by assembling the multiplier and the low-enriched uranium solution to enable the full use of neutrons that backscatter to the source direction and further improve the utilization of neutrons.A model based on SuperMC was used to evaluate the neutronics and safety behavior of the subcritical system,such as the neutron effective multiplication factor,neutron energy spectrum,medical isotope yield,and heat deposition.Based on the calculation results,when the intensity of the neutron source was 59×10^(13)n/s,the optimized design with a stacked structure could increase the yield of ^(99)Mo to182 Ci/day,which is approximately 16% higher than that obtained with a single-layer structure.The inlet H_(2)O coolant velocity of 1.0 m/s and initial temperature of 20℃ were also found to be sufficient to prevent boiling of the fuel solution.

Investigation of enhancement in planar fast neutron detector efficiency with stacked structure using Geant4

Geant4 based Monte Carlo study has been carried out to assess the improvement in efficiency of the planar structure of Silicon Carbide(SiC)-based semiconductor fast neutron detector with the stacked structure. A proton recoil detector was simulated, which consists of hydrogenous converter, i.e., high-density polyethylene(HDPE) for generating recoil protons by means of neutron elastic scattering(n, p) reaction and semiconductor material SiC, for generating a detectable electrical signal upon transport of recoil protons through it. SiC is considered in order to overcome the various factors associated with conventional Si-based devices such as operability in a harsh radiation environment, as often encountered in nuclear facilities. Converter layer thickness is optimized by considering 10~9 neutron events of different monoenergetic neutron sources as well as ^(241)Am-Be neutron spectrum. It is found that the optimized thickness for neutron energy range of 1–10 MeV is ~400 μm. However, the efficiency of fast neutron detection is estimated to be only 0.112%,which is considered very low for meaningful and reliable detection of neutrons. To overcome this problem, a stacked juxtaposition of converter layer between SiC layers has been analyzed in order to achieve high efficiency. It is noted that a tenfold efficiency improvement has been obtained—1.04% for 10 layers stacked configuration vis-à-vis 0.112% of single converter layer detector. Further simulation of the stacked detector with respect to variable converter thickness has been performed to achieve the efficiency as high as ~3.85% with up to 50 stacks.

Iterative learning-based many-objective history matching using deep neural network with stacked autoencoder

This paper presents an innovative data-integration that uses an iterative-learning method,a deep neural network(DNN)coupled with a stacked autoencoder(SAE)to solve issues encountered with many-objective history matching.The proposed method consists of a DNN-based inverse model with SAE-encoded static data and iterative updates of supervised-learning data are based on distance-based clustering schemes.DNN functions as an inverse model and results in encoded flattened data,while SAE,as a pre-trained neural network,successfully reduces dimensionality and reliably reconstructs geomodels.The iterative-learning method can improve the training data for DNN by showing the error reduction achieved with each iteration step.The proposed workflow shows the small mean absolute percentage error below 4%for all objective functions,while a typical multi-objective evolutionary algorithm fails to significantly reduce the initial population uncertainty.Iterative learning-based manyobjective history matching estimates the trends in water cuts that are not reliably included in dynamicdata matching.This confirms the proposed workflow constructs more plausible geo-models.The workflow would be a reliable alternative to overcome the less-convergent Pareto-based multi-objective evolutionary algorithm in the presence of geological uncertainty and varying objective functions.