Improving Badminton Action Recognition Using Spatio-Temporal Analysis and a Weighted Ensemble Learning Model

Incredible progress has been made in human action recognition(HAR),significantly impacting computer vision applications in sports analytics.However,identifying dynamic and complex movements in sports like badminton remains challenging due to the need for precise recognition accuracy and better management of complex motion patterns.Deep learning techniques like convolutional neural networks(CNNs),long short-term memory(LSTM),and graph convolutional networks(GCNs)improve recognition in large datasets,while the traditional machine learning methods like SVM(support vector machines),RF(random forest),and LR(logistic regression),combined with handcrafted features and ensemble approaches,perform well but struggle with the complexity of fast-paced sports like badminton.We proposed an ensemble learning model combining support vector machines(SVM),logistic regression(LR),random forest(RF),and adaptive boosting(AdaBoost)for badminton action recognition.The data in this study consist of video recordings of badminton stroke techniques,which have been extracted into spatiotemporal data.The three-dimensional distance between each skeleton point and the right hip represents the spatial features.The temporal features are the results of Fast Dynamic Time Warping(FDTW)calculations applied to 15 frames of each video sequence.The weighted ensemble model employs soft voting classifiers from SVM,LR,RF,and AdaBoost to enhance the accuracy of badminton action recognition.The E2 ensemble model,which combines SVM,LR,and AdaBoost,achieves the highest accuracy of 95.38%.

A redundant subspace weighting procedure for clock ensemble

A redundant-subspace-weighting(RSW)-based approach is proposed to enhance the frequency stability on a time scale of a clock ensemble.In this method,multiple overlapping subspaces are constructed in the clock ensemble,and the weight of each clock in this ensemble is defined by using the spatial covariance matrix.The superimposition average of covariances in different subspaces reduces the correlations between clocks in the same laboratory to some extent.After optimizing the parameters of this weighting procedure,the frequency stabilities of virtual clock ensembles are significantly improved in most cases.

Applications of Bias-removed Ensemble Mean in the Gale Forecasts over the Yellow Sea and the Bohai Sea

Based on the daily sea surface wind field prediction data of Japan Meteorological Agency(JMA) forecast model,National Centers for Environmental Prediction(NCEP GFS) model and U.S.Navy Operational Global Atmospheric Prediction System(NOGAPS) model at 12:00 UTC from June 28 to August 10 in 2009,the bias-removed ensemble mean(BRE) was used to do the forecast test on the sea surface wind fields,and the root-mean-square error(RMSE) was used to test and evaluate the forecast results.The results showed that the BRE considerably reduced the RMSEs of 24 and 48 h sea surface wind field forecasts,and the forecast skill was superior to that of the single model forecast.The RMSE decreases in the south of central Bohai Sea and the middle of the Yellow Sea were the most obvious.In addition,the BRE forecast improved evidently the forecast skill of the gale process which occurred during July 13-14 and August 7 in 2009.The forecast accuracy of the wind speed and the gale location was also improved.

Multimodel Ensemble Forecast of Global Horizontal Irradiance at PV Power Stations Based on Dynamic Variable Weight

In the present study,multimodel ensemble forecast experiments of the global horizontal irradiance(GHI)were conducted using the dynamic variable weight technique.The study was based on the forecasts of four numerical models,namely,the China Meteorological Administration Wind Energy and Solar Energy Prediction System,the Mesoscale Weather Numerical Prediction System of China Meteorological Administration,the China Meteorological Administration Regional Mesoscale Numerical Prediction System-Guangdong,and the Weather Research and Forecasting Model-Solar,and observational data from four photovoltaic(PV)power stations in Yangjiang City,Guangdong Province.The results show that compared with those of the monthly optimal numerical model forecasts,the dynamic variable weight-based ensemble forecasts exhibited 0.97%-15.96%smaller values of the mean absolute error and 3.31%-18.40%lower values of the root mean square error(RMSE).However,the increase in the correlation coefficient was not obvious.Specifically,the multimodel ensemble mainly improved the performance of GHI forecasts below 700 W m^(-2),particularly below 400 W m^(-2),with RMSE reductions as high as 7.56%-28.28%.In contrast,the RMSE increased at GHI levels above 700 W m^(-2).As for the key period of PV power station output(02:00-07:00),the accuracy of GHI forecasts could be improved by the multimodel ensemble:the multimodel ensemble could effectively decrease the daily maximum absolute error(AE max)of GHI forecasts.Moreover,with increasing forecasting difficulty under cloudy conditions,the multimodel ensemble,which yields data closer to the actual observations,could simulate GHI fluctuations more accurately.

Assimilating satellite SST/SSH and in-situ T/S profiles with the Localized Weighted Ensemble Kalman Filter

The Localized Weighted Ensemble Kalman Filter(LWEnKF)is a new nonlinear/non-Gaussian data assimilation(DA)method that can effectively alleviate the filter degradation problem faced by particle filtering,and it has great prospects for applications in geophysical models.In terms of operational applications,along-track sea surface height(AT-SSH),swath sea surface temperature(S-SST)and in-situ temperature and salinity(T/S)profiles are assimilated using the LWEnKF in the northern South China Sea(SCS).To adapt to the vertical S-coordinates of the Regional Ocean Modelling System(ROMS),a vertical localization radius function is designed for T/S profiles assimilation using the LWEnKF.The results show that the LWEnKF outperforms the local particle filter(LPF)due to the introduction of the Ensemble Kalman Filter(EnKF)as a proposal density;the RMSEs of SSH and SST from the LWEnKF are comparable to the EnKF,but the RMSEs of T/S profiles reduce significantly by approximately 55%for the T profile and 35%for the S profile(relative to the EnKF).As a result,the LWEnKF makes more reasonable predictions of the internal ocean temperature field.In addition,the three-dimensional structures of nonlinear mesoscale eddies are better characterized when using the LWEnKF.

Game Theory-Based Dynamic Weighted Ensemble for Retinal Disease Classification

An automated retinal disease detection system has long been in exis-tence and it provides a safe,no-contact and cost-effective solution for detecting this disease.This paper presents a game theory-based dynamic weighted ensem-ble of a feature extraction-based machine learning model and a deep transfer learning model for automatic retinal disease detection.The feature extraction-based machine learning model uses Gaussian kernel-based fuzzy rough sets for reduction of features,and XGBoost classifier for the classification.The transfer learning model uses VGG16 or ResNet50 or Inception-ResNet-v2.A novel ensemble classifier based on the game theory approach is proposed for the fusion of the outputs of the transfer learning model and the XGBoost classifier model.The ensemble approach significantly improves the accuracy of retinal disease pre-diction and results in an excellent performance when compared to the individual deep learning and feature-based models.

Ensemble Based Temporal Weighting and Pareto Ranking (ETP) Model for Effective Root Cause Analysis

Root-cause identification plays a vital role in business decision making by providing effective future directions for the organizations.Aspect extraction and sentiment extraction plays a vital role in identifying the rootcauses.This paper proposes the Ensemble based temporal weighting and pareto ranking(ETP)model for Root-cause identification.Aspect extraction is performed based on rules and is followed by opinion identification using the proposed boosted ensemble model.The obtained aspects are validated and ranked using the proposed aspect weighing scheme.Pareto-rule based aspect selection is performed as the final selection mechanism and the results are presented for business decision making.Experiments were performed with the standard five product benchmark dataset.Performances on all five product reviews indicate the effective performance of the proposed model.Comparisons are performed using three standard state-of-the-art models and effectiveness is measured in terms of F-Measure and Detection rates.The results indicate improved performances exhibited by the proposed model with an increase in F-Measure levels at 1%–15%and detection rates at 4%–24%compared to the state-of-the-art models.

Optimization Ensemble Weights Model for Wind Forecasting System

Effective technology for wind direction forecasting can be realized using the recent advances in machine learning.Consequently,the stability and safety of power systems are expected to be significantly improved.However,the unstable and unpredictable qualities of the wind predict the wind direction a challenging problem.This paper proposes a practical forecasting approach based on the weighted ensemble of machine learning models.This weighted ensemble is optimized using a whale optimization algorithm guided by particle swarm optimization(PSO-Guided WOA).The proposed optimized weighted ensemble predicts the wind direction given a set of input features.The conducted experiments employed the wind power forecasting dataset,freely available on Kaggle and developed to predict the regular power generation at seven wind farms over forty-eight hours.The recorded results of the conducted experiments emphasize the effectiveness of the proposed ensemble in achieving accurate predictions of the wind direction.In addition,a comparison is established between the proposed optimized ensemble and other competing optimized ensembles to prove its superiority.Moreover,statistical analysis using one-way analysis of variance(ANOVA)and Wilcoxon’s rank-sum are provided based on the recorded results to confirm the excellent accuracy achieved by the proposed optimized weighted ensemble.

Neural Network Compact Ensemble and Its Applications

There has been many methods in constructing neural network （NN） ensembles, where the method of simultaneous training has succeed in generalization performance and efficiency. But just like regular methods of constructing NN ensembles, it follows the two steps, first training component networks, and then combining them. As the two steps being independent, an assumption is used to facilitate interactions among NNs during the training stage. This paper presents a compact ensemble method which integrates the two steps of ensemble construction into one step by attempting to train individual NNs in an ensemble and weigh the individual members adaptively according to their individual performance in the same learning process. This provides an opportunity for the individual NNs to interact with each other based on their real contributions to the ensemble. The classification performance of NN compact ensemble （NNCE） was validated through some benchmark problems in machine learning, including Australian credit card assessment, pima Indians diabetes, heart disease, breast cancer and glass. Compared with other ensembles, the classification error rate of NNCE can be decreased by 0.45% to 68%. In addition, the NNCE was applied to fault diagnosis for rolling element bearing. The 11 time-domain statistical features are extracted as the properties of data, and the NNCE is employed to classify the data. With the results of several experiments, the compact ensemble method is shown to give good generalization performance. The compact ensemble method can recognize the different fault types and various fault degrees of the same fault type.

An Intelligent Hazardous Waste Detection and Classification Model Using Ensemble Learning Techniques

Proper waste management models using recent technologies like computer vision,machine learning(ML),and deep learning(DL)are needed to effectively handle the massive quantity of increasing waste.Therefore,waste classification becomes a crucial topic which helps to categorize waste into hazardous or non-hazardous ones and thereby assist in the decision making of the waste management process.This study concentrates on the design of hazardous waste detection and classification using ensemble learning(HWDC-EL)technique to reduce toxicity and improve human health.The goal of the HWDC-EL technique is to detect the multiple classes of wastes,particularly hazardous and non-hazardous wastes.The HWDC-EL technique involves the ensemble of three feature extractors using Model Averaging technique namely discrete local binary patterns(DLBP),EfficientNet,and DenseNet121.In addition,the flower pollination algorithm(FPA)based hyperparameter optimizers are used to optimally adjust the parameters involved in the EfficientNet and DenseNet121 models.Moreover,a weighted voting-based ensemble classifier is derived using three machine learning algorithms namely support vector machine(SVM),extreme learning machine(ELM),and gradient boosting tree(GBT).The performance of the HWDC-EL technique is tested using a benchmark Garbage dataset and it obtains a maximum accuracy of 98.85%.

Ensemble Nonlinear Support Vector Machine Approach for Predicting Chronic Kidney Diseases

Urban living in large modern cities exerts considerable adverse effectson health and thus increases the risk of contracting several chronic kidney diseases (CKD). The prediction of CKDs has become a major task in urbanizedcountries. The primary objective of this work is to introduce and develop predictive analytics for predicting CKDs. However, prediction of huge samples isbecoming increasingly difficult. Meanwhile, MapReduce provides a feasible framework for programming predictive algorithms with map and reduce functions.The relatively simple programming interface helps solve problems in the scalability and efficiency of predictive learning algorithms. In the proposed work, theiterative weighted map reduce framework is introduced for the effective management of large dataset samples. A binary classification problem is formulated usingensemble nonlinear support vector machines and random forests. Thus, instead ofusing the normal linear combination of kernel activations, the proposed work creates nonlinear combinations of kernel activations in prototype examples. Furthermore, different descriptors are combined in an ensemble of deep support vectormachines, where the product rule is used to combine probability estimates ofdifferent classifiers. Performance is evaluated in terms of the prediction accuracyand interpretability of the model and the results.

事件触发机制下配电网三相动态状态估计

随着高级量测体系的发展和智能电表的广泛应用,为配电网三相状态估计提供了丰富的终端量测信息。与此同时,大量的智能电表数据给配电网通信系统提出了更高的通信带宽和实时存储要求。为了缓解量测拥堵和时延现象,文中引入事件触发机制代替传统量测数据的周期性采样,在保证有效量测信息及时上传的同时减少通信成本和投资。在此基础上,针对配电网实时状态感知问题,提出了基于鲁棒集合卡尔曼滤波的配电网三相动态状态估计方法,在正常运行场景下,能够保持与无偏估计加权最小二乘法相近的估计精度;当含有坏数据时,该方法也拥有较强的鲁棒性。

基于ER Rule的多分类器汽车评论情感分类研究

该文针对汽车评论语料的情感二分类问题,提出一种基于证据推理规则的多分类器融合的情感分类方法。在情感特征构建方面,通过实验对比不同特征模型对分类结果的影响,并改进传统的TFIDF权重计算方法。同时,在此基础上使用ER Rule融合不同分类器进行文本情感极性分析,并考虑各分类器的权重和可靠度。最后,爬取汽车网站上的评论数据对上述方法进行测试,并用公开的中文酒店评论语料数据进行了验证,结果表明该方法能够有效集成不同分类器的优点,与传统机器学习分类算法相比,其结果在Recall,F1值和Accuracy三个指标上得到了提高,与目前流行的深度学习算法和集成学习算法相比,其结果总体占优。

基于特征提取和最优加权集成策略的风机叶片结冰故障检测

针对风机叶片结冰检测中现有集成方法不能充分发挥不同个体分类器优势的问题,提出了一种基于特征提取和最优加权集成学习的叶片结冰检测模型。首先,用堆叠降噪自动编码器提取结冰关联特征后,考虑不同单一分类器在二分类应用中的表现及其差异,选择随机森林、极限梯度提升树、轻量梯度提升机、K-近邻算法作为个体学习器,并用贝叶斯算法对其进行超参数优化。然后提出基于序列二次规划的最优加权集成策略对叶片状态进行判别。最后利用金风科技提供的15号和21号风机的历史数据进行了仿真实验,结果表明:所提出的检测模型与个体学习器及其他集成模型相比多项指标均有所提升,准确度达到了99.2%,在结冰检测方面具有一定的有效性。

基于特征判定系数的电力变压器振动信号故障诊断

变压器带电故障诊断对于保证电力变压器安全平稳运行具有重要的意义。针对变压器工作环境复杂且单一参数表征变压器故障类型不全面的问题,文中提出一种基于自适应噪声完备集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)和特征熵权法(entropy weight method,EWM)进行故障诊断的方法。通过相关系数与峭度加权(correlation coefficient and weighted kurtosis,CCWK)原则筛选CEEMDAN分量并重构信号,在实现剔除冗余分量的同时,提升变压器振动信号特征的表征能力;利用EWM构建特征判定系数实现单一数据诊断变压器故障类型;通过主成分分析法减小混合域特征尺度,采用鸡群优化算法优化支持向量机(support vector machine,SVM)模型进行故障诊断。对某变电站110 kV三相油浸式变压器进行分析,结果表明与概率神经网络和SVM等变压器故障诊断方法相比,文中方法能在提前定性故障类型的同时,进一步提高变压器故障诊断的准确率与效率。

CEEMDAN-WPE-CLSA超短期风电功率预测方法研究

提出了一种结合自适应噪声完全集合经验模态分解、加权排列熵、卷积神经网络、长短期记忆网络和自注意力机制的超短期风电功率预测方法。首先,利用自适应噪声完全集合经验模态分解将原始风电功率时间序列自适应分解为一系列的模态分量,降低原始序列的非线性和波动性;其次,根据加权排列熵计算各模态分量间的相似性并对相似的分量进行重组,以修正自适应噪声完全集合经验模态分解的过度分解问题,使得修正后的模态分量更具规律性;最后,将重组后的分量输入卷积长短期记忆网络进行时序建模,并利用自注意力机制对卷积长短期记忆网络的神经元权重进行重新分配,提高了卷积长短期记忆网络对输入特征不确定性的适应能力。在此基础上,明确了自注意力机制和自适应噪声完全集合经验模态分解、加权排列熵在风电功率预测中的作用机制,以及风电功率信号包含的重要物理信息,证明了自适应噪声完全集合经验模态分解、加权排列熵以及自注意力机制在风电功率信号模态分解和长短期记忆网络隐层输出权重分配中的有效性。

一种三层加权文本聚类集成方法

为了提高聚类集成效果,本文设计了一种对点、簇、划分进行加权的统一框架,提出一种三层加权文本聚类集成方法。首先根据基聚类生成超图邻接矩阵,然后依次对点、簇、划分进行加权获得加权邻接矩阵,最后用层次凝聚聚类算法获得最终结果。在多个真实文本数据集上进行实验,结果表明,与未加权及其他层面加权相比,三层加权方法可以获得更好的聚类效果,三层加权相较于未加权的平均提升幅度为12.02%;与近年来的其他8种加权方法相比,该方法在所有数据集上的平均排名位列第一,验证了本文方法的有效性。

基于增量加权的不平衡漂移数据流分类算法

概念漂移是数据流学习领域中的一个难点问题,同时数据流中存在的类不平衡问题也会严重影响算法的分类性能。针对概念漂移和类不平衡的联合问题,在基于数据块集成的方法上引入在线更新机制,结合重采样和遗忘机制提出了一种增量加权集成的不平衡数据流分类方法(incremental weighted ensemble for imbalance learning,IWEIL)。该方法以集成框架为基础,利用基于可变大小窗口的遗忘机制确定基分类器对窗口内最近若干实例的分类性能,并计算基分类器的权重,随着新实例的逐个到达,在线更新IWEIL中每个基分器及其权重。同时,使用改进的自适应最近邻SMOTE方法生成符合新概念的新少数类实例以解决数据流中类不平衡问题。在人工数据集和真实数据集上进行实验,结果表明,相比于DWMIL算法,IWEIL在HyperPlane数据集上的G-mean和recall指标分别提升了5.77%和6.28%,在Electricity数据集上两个指标分别提升了3.25%和6.47%。最后,IWEIL在安卓应用检测问题上表现良好。

中国冬季降水的支持向量机预测模型研究

我国冬季降水对于农业、水资源管理和自然灾害风险评估具有重要意义.受多种气象因素的影响,冬季降水的预测仍具有挑战性,进一步提升冬季降水的预测技巧是当下短期气候预测研究的重要课题.本研究采用支持向量机(SVM)方法,旨在通过机器学习方法提高中国冬季降水的预测准确率.基于NCEP_CFS, ECMWF_SYSTEM, BCC_CSM等五个模式数据以及站点数据,建立针对冬季降水的SVM集成预测模型,并与单个模式和等权集合平均模型(AVE)加以对比.SVM模型因其强泛化和处理非线性问题的能力,在中国冬季降水预测中表现良好.研究表明:(1)SVM模型较单个模式及AVE模型的预测准确性与稳定性得到大幅提升,SVM模型的PS评分和PCS评分显著高于单个成员模式的结果,最大分别提高了8.0(12.6%)和3.9(7.4%),较AVE模型则最大分别提高了5.4(8.2%)和2.1(3.8%),预报技巧的提高在观测资料相对缺乏的西南和西北地区尤为明显.(2)从均方根误差和时间相关系数的空间分布上来看,SVM模型对其成员模式在西藏地区、西南地区、华东及华南地区误差较大的情况改善明显,误差最大降低了259(90.9%),预报技巧最大提高了1.13.(3)独立样本检验中,SVM模型的PS评分和PCS评分显著高于单个模式和AVE模型,最大提高了10.79(20.3%)和11.39(27.3%).因此,SVM模型的构建,将有助于进一步提高中国冬季降水预测的准确性和稳定性,为气象防灾减灾和气候资源开发利用等提供重要技术支撑.

城市土地价格时空预测Stacking-GWR模型

城市土地价格影响国土空间规划决策、现代城市治理和土地市场调控,预测城市土地价格具有重要意义,但不同用途的土地价格变化趋势差异显著且具有空间异质性,很难用单个模型进行预测。该文提出一种城市土地价格时空预测Stacking-GWR模型,以常州市主城区为研究区,根据土地价格变化趋势分为工业用地和非工业用地两组,利用Stacking-GWR模型进行土地价格预测,并与单独使用Stacking、地理加权回归(GWR)、时空地理加权回归(GTWR)模型的预测结果进行对比分析。结果表明:①Stacking-GWR模型融合了地价数据中的特征、空间和时间信息,能提高预测精度;②根据土地价格变化趋势进行分组后,模型预测精度优于不分组时的预测精度;③工业用地和非工业用地土地价格的全局和邻域影响因子差异显著。