基于Blending-Clustering集成学习的大坝变形预测模型

【目的】变形是反映大坝结构性态最直观的效应量,构建科学合理的变形预测模型是保障大坝安全健康运行的重要手段。针对传统大坝变形预测模型预测精度低、误报率高等问题导致的错误报警现象,【方法】选取不同预测模型和聚类算法集成,构建了一种Blending-Clustering集成学习的大坝变形预测模型,该模型以Blending对单一预测模型集成提升预测精度为核心,并通过Clustering聚类优选预测值改善模型稳定性。以新疆某面板堆石坝变形监测数据为实例分析,通过多模型预测性能比较,对所提出模型的预测精度和稳定性进行全面评估。【结果】结果显示:Blending-Clustering模型将预测模型和聚类算法集成,均方根误差(RMSE)和归一化平均百分比误差(nMAPE)明显降低,模型的预测精度得到显著提高;回归相关系数(R~2)得到提升,模型具备更强的拟合能力;在面板堆石坝上22个测点变形数据集上的预测评价指标波动范围更小,模型的泛化性和稳定性得到有效增强。【结论】结果表明:Blending-Clustering集成预测模型对于预测精度、泛化性和稳定性均有明显提升,在实际工程具有一定的应用价值。

Hierarchical multihead self-attention for time-series-based fault diagnosis

Fault diagnosis is important for maintaining the safety and effectiveness of chemical process.Considering the multivariate,nonlinear,and dynamic characteristic of chemical process,many time-series-based data-driven fault diagnosis methods have been developed in recent years.However,the existing methods have the problem of long-term dependency and are difficult to train due to the sequential way of training.To overcome these problems,a novel fault diagnosis method based on time-series and the hierarchical multihead self-attention(HMSAN)is proposed for chemical process.First,a sliding window strategy is adopted to construct the normalized time-series dataset.Second,the HMSAN is developed to extract the time-relevant features from the time-series process data.It improves the basic self-attention model in both width and depth.With the multihead structure,the HMSAN can pay attention to different aspects of the complicated chemical process and obtain the global dynamic features.However,the multiple heads in parallel lead to redundant information,which cannot improve the diagnosis performance.With the hierarchical structure,the redundant information is reduced and the deep local time-related features are further extracted.Besides,a novel many-to-one training strategy is introduced for HMSAN to simplify the training procedure and capture the long-term dependency.Finally,the effectiveness of the proposed method is demonstrated by two chemical cases.The experimental results show that the proposed method achieves a great performance on time-series industrial data and outperforms the state-of-the-art approaches.

Modeling urban redevelopment:A novel approach using time-series remote sensing data and machine learning

Accurate mapping and timely monitoring of urban redevelopment are pivotal for urban studies and decisionmakers to foster sustainable urban development.Traditional mapping methods heavily depend on field surveys and subjective questionnaires,yielding less objective,reliable,and timely data.Recent advancements in Geographic Information Systems(GIS)and remote-sensing technologies have improved the identification and mapping of urban redevelopment through quantitative analysis using satellite-based observations.Nonetheless,challenges persist,particularly concerning accuracy and significant temporal delays.This study introduces a novel approach to modeling urban redevelopment,leveraging machine learning algorithms and remote-sensing data.This methodology can facilitate the accurate and timely identification of urban redevelopment activities.The study’s machine learning model can analyze time-series remote-sensing data to identify spatio-temporal and spectral patterns related to urban redevelopment.The model is thoroughly evaluated,and the results indicate that it can accurately capture the time-series patterns of urban redevelopment.This research’s findings are useful for evaluating urban demographic and economic changes,informing policymaking and urban planning,and contributing to sustainable urban development.The model can also serve as a foundation for future research on early-stage urban redevelopment detection and evaluation of the causes and impacts of urban redevelopment.

Missing Value Imputation for Radar-Derived Time-Series Tracks of Aerial Targets Based on Improved Self-Attention-Based Network

The frequent missing values in radar-derived time-series tracks of aerial targets(RTT-AT)lead to significant challenges in subsequent data-driven tasks.However,the majority of imputation research focuses on random missing(RM)that differs significantly from common missing patterns of RTT-AT.The method for solving the RM may experience performance degradation or failure when applied to RTT-AT imputation.Conventional autoregressive deep learning methods are prone to error accumulation and long-term dependency loss.In this paper,a non-autoregressive imputation model that addresses the issue of missing value imputation for two common missing patterns in RTT-AT is proposed.Our model consists of two probabilistic sparse diagonal masking self-attention(PSDMSA)units and a weight fusion unit.It learns missing values by combining the representations outputted by the two units,aiming to minimize the difference between the missing values and their actual values.The PSDMSA units effectively capture temporal dependencies and attribute correlations between time steps,improving imputation quality.The weight fusion unit automatically updates the weights of the output representations from the two units to obtain a more accurate final representation.The experimental results indicate that,despite varying missing rates in the two missing patterns,our model consistently outperforms other methods in imputation performance and exhibits a low frequency of deviations in estimates for specific missing entries.Compared to the state-of-the-art autoregressive deep learning imputation model Bidirectional Recurrent Imputation for Time Series(BRITS),our proposed model reduces mean absolute error(MAE)by 31%~50%.Additionally,the model attains a training speed that is 4 to 8 times faster when compared to both BRITS and a standard Transformer model when trained on the same dataset.Finally,the findings from the ablation experiments demonstrate that the PSDMSA,the weight fusion unit,cascade network design,and imputation loss enhance imputation performance and confirm the efficacy of our design.

A Shared Natural Neighbors Based-Hierarchical Clustering Algorithm for Discovering Arbitrary-Shaped Clusters

In clustering algorithms,the selection of neighbors significantly affects the quality of the final clustering results.While various neighbor relationships exist,such as K-nearest neighbors,natural neighbors,and shared neighbors,most neighbor relationships can only handle single structural relationships,and the identification accuracy is low for datasets with multiple structures.In life,people’s first instinct for complex things is to divide them into multiple parts to complete.Partitioning the dataset into more sub-graphs is a good idea approach to identifying complex structures.Taking inspiration from this,we propose a novel neighbor method:Shared Natural Neighbors(SNaN).To demonstrate the superiority of this neighbor method,we propose a shared natural neighbors-based hierarchical clustering algorithm for discovering arbitrary-shaped clusters(HC-SNaN).Our algorithm excels in identifying both spherical clusters and manifold clusters.Tested on synthetic datasets and real-world datasets,HC-SNaN demonstrates significant advantages over existing clustering algorithms,particularly when dealing with datasets containing arbitrary shapes.

Multiscale and Auto-Tuned Semi-Supervised Deep Subspace Clustering and Its Application in Brain Tumor Clustering

In this paper,we introduce a novel Multi-scale and Auto-tuned Semi-supervised Deep Subspace Clustering(MAS-DSC)algorithm,aimed at addressing the challenges of deep subspace clustering in high-dimensional real-world data,particularly in the field of medical imaging.Traditional deep subspace clustering algorithms,which are mostly unsupervised,are limited in their ability to effectively utilize the inherent prior knowledge in medical images.Our MAS-DSC algorithm incorporates a semi-supervised learning framework that uses a small amount of labeled data to guide the clustering process,thereby enhancing the discriminative power of the feature representations.Additionally,the multi-scale feature extraction mechanism is designed to adapt to the complexity of medical imaging data,resulting in more accurate clustering performance.To address the difficulty of hyperparameter selection in deep subspace clustering,this paper employs a Bayesian optimization algorithm for adaptive tuning of hyperparameters related to subspace clustering,prior knowledge constraints,and model loss weights.Extensive experiments on standard clustering datasets,including ORL,Coil20,and Coil100,validate the effectiveness of the MAS-DSC algorithm.The results show that with its multi-scale network structure and Bayesian hyperparameter optimization,MAS-DSC achieves excellent clustering results on these datasets.Furthermore,tests on a brain tumor dataset demonstrate the robustness of the algorithm and its ability to leverage prior knowledge for efficient feature extraction and enhanced clustering performance within a semi-supervised learning framework.

Knowledge-Driven Possibilistic Clustering with Automatic Cluster Elimination

Traditional Fuzzy C-Means(FCM)and Possibilistic C-Means(PCM)clustering algorithms are data-driven,and their objective function minimization process is based on the available numeric data.Recently,knowledge hints have been introduced to formknowledge-driven clustering algorithms,which reveal a data structure that considers not only the relationships between data but also the compatibility with knowledge hints.However,these algorithms cannot produce the optimal number of clusters by the clustering algorithm itself;they require the assistance of evaluation indices.Moreover,knowledge hints are usually used as part of the data structure(directly replacing some clustering centers),which severely limits the flexibility of the algorithm and can lead to knowledgemisguidance.To solve this problem,this study designs a newknowledge-driven clustering algorithmcalled the PCM clusteringwith High-density Points(HP-PCM),in which domain knowledge is represented in the form of so-called high-density points.First,a newdatadensitycalculation function is proposed.The Density Knowledge Points Extraction(DKPE)method is established to filter out high-density points from the dataset to form knowledge hints.Then,these hints are incorporated into the PCM objective function so that the clustering algorithm is guided by high-density points to discover the natural data structure.Finally,the initial number of clusters is set to be greater than the true one based on the number of knowledge hints.Then,the HP-PCM algorithm automatically determines the final number of clusters during the clustering process by considering the cluster elimination mechanism.Through experimental studies,including some comparative analyses,the results highlight the effectiveness of the proposed algorithm,such as the increased success rate in clustering,the ability to determine the optimal cluster number,and the faster convergence speed.

A novel method for clustering cellular data to improve classification

Many fields,such as neuroscience,are experiencing the vast prolife ration of cellular data,underscoring the need fo r organizing and interpreting large datasets.A popular approach partitions data into manageable subsets via hierarchical clustering,but objective methods to determine the appropriate classification granularity are missing.We recently introduced a technique to systematically identify when to stop subdividing clusters based on the fundamental principle that cells must differ more between than within clusters.Here we present the corresponding protocol to classify cellular datasets by combining datadriven unsupervised hierarchical clustering with statistical testing.These general-purpose functions are applicable to any cellular dataset that can be organized as two-dimensional matrices of numerical values,including molecula r,physiological,and anatomical datasets.We demonstrate the protocol using cellular data from the Janelia MouseLight project to chara cterize morphological aspects of neurons.

Path-Based Clustering Algorithm with High Scalability Using the Combined Behavior of Evolutionary Algorithms

Path-based clustering algorithms typically generate clusters by optimizing a benchmark function.Most optimiza-tion methods in clustering algorithms often offer solutions close to the general optimal value.This study achieves the global optimum value for the criterion function in a shorter time using the minimax distance,Maximum Spanning Tree“MST”,and meta-heuristic algorithms,including Genetic Algorithm“GA”and Particle Swarm Optimization“PSO”.The Fast Path-based Clustering“FPC”algorithm proposed in this paper can find cluster centers correctly in most datasets and quickly perform clustering operations.The FPC does this operation using MST,the minimax distance,and a new hybrid meta-heuristic algorithm in a few rounds of algorithm iterations.This algorithm can achieve the global optimal value,and the main clustering process of the algorithm has a computational complexity of O�k2×n�.However,due to the complexity of the minimum distance algorithm,the total computational complexity is O�n2�.Experimental results of FPC on synthetic datasets with arbitrary shapes demonstrate that the algorithm is resistant to noise and outliers and can correctly identify clusters of varying sizes and numbers.In addition,the FPC requires the number of clusters as the only parameter to perform the clustering process.A comparative analysis of FPC and other clustering algorithms in this domain indicates that FPC exhibits superior speed,stability,and performance.

Efficient Clustering Network Based on Matrix Factorization

Contrastive learning is a significant research direction in the field of deep learning.However,existing data augmentation methods often lead to issues such as semantic drift in generated views while the complexity of model pre-training limits further improvement in the performance of existing methods.To address these challenges,we propose the Efficient Clustering Network based on Matrix Factorization(ECN-MF).Specifically,we design a batched low-rank Singular Value Decomposition(SVD)algorithm for data augmentation to eliminate redundant information and uncover major patterns of variation and key information in the data.Additionally,we design a Mutual Information-Enhanced Clustering Module(MI-ECM)to accelerate the training process by leveraging a simple architecture to bring samples from the same cluster closer while pushing samples from other clusters apart.Extensive experiments on six datasets demonstrate that ECN-MF exhibits more effective performance compared to state-of-the-art algorithms.

Improved Data Stream Clustering Method: Incorporating KD-Tree for Typicality and Eccentricity-Based Approach

Data stream clustering is integral to contemporary big data applications.However,addressing the ongoing influx of data streams efficiently and accurately remains a primary challenge in current research.This paper aims to elevate the efficiency and precision of data stream clustering,leveraging the TEDA(Typicality and Eccentricity Data Analysis)algorithm as a foundation,we introduce improvements by integrating a nearest neighbor search algorithm to enhance both the efficiency and accuracy of the algorithm.The original TEDA algorithm,grounded in the concept of“Typicality and Eccentricity Data Analytics”,represents an evolving and recursive method that requires no prior knowledge.While the algorithm autonomously creates and merges clusters as new data arrives,its efficiency is significantly hindered by the need to traverse all existing clusters upon the arrival of further data.This work presents the NS-TEDA(Neighbor Search Based Typicality and Eccentricity Data Analysis)algorithm by incorporating a KD-Tree(K-Dimensional Tree)algorithm integrated with the Scapegoat Tree.Upon arrival,this ensures that new data points interact solely with clusters in very close proximity.This significantly enhances algorithm efficiency while preventing a single data point from joining too many clusters and mitigating the merging of clusters with high overlap to some extent.We apply the NS-TEDA algorithm to several well-known datasets,comparing its performance with other data stream clustering algorithms and the original TEDA algorithm.The results demonstrate that the proposed algorithm achieves higher accuracy,and its runtime exhibits almost linear dependence on the volume of data,making it more suitable for large-scale data stream analysis research.

Hyperspectral Image Based Interpretable Feature Clustering Algorithm

Hyperspectral imagery encompasses spectral and spatial dimensions,reflecting the material properties of objects.Its application proves crucial in search and rescue,concealed target identification,and crop growth analysis.Clustering is an important method of hyperspectral analysis.The vast data volume of hyperspectral imagery,coupled with redundant information,poses significant challenges in swiftly and accurately extracting features for subsequent analysis.The current hyperspectral feature clustering methods,which are mostly studied from space or spectrum,do not have strong interpretability,resulting in poor comprehensibility of the algorithm.So,this research introduces a feature clustering algorithm for hyperspectral imagery from an interpretability perspective.It commences with a simulated perception process,proposing an interpretable band selection algorithm to reduce data dimensions.Following this,amulti-dimensional clustering algorithm,rooted in fuzzy and kernel clustering,is developed to highlight intra-class similarities and inter-class differences.An optimized P systemis then introduced to enhance computational efficiency.This system coordinates all cells within a mapping space to compute optimal cluster centers,facilitating parallel computation.This approach diminishes sensitivity to initial cluster centers and augments global search capabilities,thus preventing entrapment in local minima and enhancing clustering performance.Experiments conducted on 300 datasets,comprising both real and simulated data.The results show that the average accuracy(ACC)of the proposed algorithm is 0.86 and the combination measure(CM)is 0.81.

Deep Learning and Tensor-Based Multiple Clustering Approaches for Cyber-Physical-Social Applications

The study delves into the expanding role of network platforms in our daily lives, encompassing various mediums like blogs, forums, online chats, and prominent social media platforms such as Facebook, Twitter, and Instagram. While these platforms offer avenues for self-expression and community support, they concurrently harbor negative impacts, fostering antisocial behaviors like phishing, impersonation, hate speech, cyberbullying, cyberstalking, cyberterrorism, fake news propagation, spamming, and fraud. Notably, individuals also leverage these platforms to connect with authorities and seek aid during disasters. The overarching objective of this research is to address the dual nature of network platforms by proposing innovative methodologies aimed at enhancing their positive aspects and mitigating their negative repercussions. To achieve this, the study introduces a weight learning method grounded in multi-linear attribute ranking. This approach serves to evaluate the significance of attribute combinations across all feature spaces. Additionally, a novel clustering method based on tensors is proposed to elevate the quality of clustering while effectively distinguishing selected features. The methodology incorporates a weighted average similarity matrix and optionally integrates weighted Euclidean distance, contributing to a more nuanced understanding of attribute importance. The analysis of the proposed methods yields significant findings. The weight learning method proves instrumental in discerning the importance of attribute combinations, shedding light on key aspects within feature spaces. Simultaneously, the clustering method based on tensors exhibits improved efficacy in enhancing clustering quality and feature distinction. This not only advances our understanding of attribute importance but also paves the way for more nuanced data analysis methodologies. In conclusion, this research underscores the pivotal role of network platforms in contemporary society, emphasizing their potential for both positive contributions and adverse consequences. The proposed methodologies offer novel approaches to address these dualities, providing a foundation for future research and practical applications. Ultimately, this study contributes to the ongoing discourse on optimizing the utility of network platforms while minimizing their negative impacts.

Examining the Use of Scott’s Formula and Link Expiration Time Metric for Vehicular Clustering

Implementing machine learning algorithms in the non-conducive environment of the vehicular network requires some adaptations due to the high computational complexity of these algorithms.K-clustering algorithms are simplistic,with fast performance and relative accuracy.However,their implementation depends on the initial selection of clusters number(K),the initial clusters’centers,and the clustering metric.This paper investigated using Scott’s histogram formula to estimate the K number and the Link Expiration Time(LET)as a clustering metric.Realistic traffic flows were considered for three maps,namely Highway,Traffic Light junction,and Roundabout junction,to study the effect of road layout on estimating the K number.A fast version of the PAM algorithm was used for clustering with a modification to reduce time complexity.The Affinity propagation algorithm sets the baseline for the estimated K number,and the Medoid Silhouette method is used to quantify the clustering.OMNET++,Veins,and SUMO were used to simulate the traffic,while the related algorithms were implemented in Python.The Scott’s formula estimation of the K number only matched the baseline when the road layout was simple.Moreover,the clustering algorithm required one iteration on average to converge when used with LET.

Design and construction of charged-particle telescope array for study of exotic nuclear clustering structure

The exploration of exotic shapes and properties of atomic nuclei,e.g.,αcluster and toroidal shape,is a fascinating field in nuclear physics.To study the decay of these nuclei,a novel detector aimed at detecting multipleα-particle events was designed and constructed.The detector comprises two layers of double-sided silicon strip detectors(DSSD)and a cesium iodide scintillator array coupled with silicon photomultipliers array as light sensors,which has the advantages of their small size,fast response,and large dynamic range.DSSDs coupled with cesium iodide crystal arrays are used to distinguish multipleαhits.The detector array has a compact and integrated design that can be adapted to different experimental conditions.The detector array was simulated using Geant4,and the excitation energy spectra of someα-clustering nuclei were reconstructed to demonstrate the performance.The simulation results show that the detector array has excellent angular and energy resolutions,enabling effective reconstruction of the nuclear excited state by multipleαparticle events.This detector offers a new and powerful tool for nuclear physics experiments and has the potential to discover interesting physical phenomena related to exotic nuclear structures and their decay mechanisms.

Density Clustering Algorithm Based on KD-Tree and Voting Rules

Traditional clustering algorithms often struggle to produce satisfactory results when dealing with datasets withuneven density. Additionally, they incur substantial computational costs when applied to high-dimensional datadue to calculating similarity matrices. To alleviate these issues, we employ the KD-Tree to partition the dataset andcompute the K-nearest neighbors (KNN) density for each point, thereby avoiding the computation of similaritymatrices. Moreover, we apply the rules of voting elections, treating each data point as a voter and casting a votefor the point with the highest density among its KNN. By utilizing the vote counts of each point, we develop thestrategy for classifying noise points and potential cluster centers, allowing the algorithm to identify clusters withuneven density and complex shapes. Additionally, we define the concept of “adhesive points” between two clustersto merge adjacent clusters that have similar densities. This process helps us identify the optimal number of clustersautomatically. Experimental results indicate that our algorithm not only improves the efficiency of clustering butalso increases its accuracy.

Sparse Reconstructive Evidential Clustering for Multi-View Data

Although many multi-view clustering(MVC) algorithms with acceptable performances have been presented, to the best of our knowledge, nearly all of them need to be fed with the correct number of clusters. In addition, these existing algorithms create only the hard and fuzzy partitions for multi-view objects,which are often located in highly-overlapping areas of multi-view feature space. The adoption of hard and fuzzy partition ignores the ambiguity and uncertainty in the assignment of objects, likely leading to performance degradation. To address these issues, we propose a novel sparse reconstructive multi-view evidential clustering algorithm(SRMVEC). Based on a sparse reconstructive procedure, SRMVEC learns a shared affinity matrix across views, and maps multi-view objects to a 2-dimensional humanreadable chart by calculating 2 newly defined mathematical metrics for each object. From this chart, users can detect the number of clusters and select several objects existing in the dataset as cluster centers. Then, SRMVEC derives a credal partition under the framework of evidence theory, improving the fault tolerance of clustering. Ablation studies show the benefits of adopting the sparse reconstructive procedure and evidence theory. Besides,SRMVEC delivers effectiveness on benchmark datasets by outperforming some state-of-the-art methods.

An air combat maneuver pattern extraction based on time series segmentation and clustering analysis

Target maneuver recognition is a prerequisite for air combat situation awareness,trajectory prediction,threat assessment and maneuver decision.To get rid of the dependence of the current target maneuver recognition method on empirical criteria and sample data,and automatically and adaptively complete the task of extracting the target maneuver pattern,in this paper,an air combat maneuver pattern extraction based on time series segmentation and clustering analysis is proposed by combining autoencoder,G-G clustering algorithm and the selective ensemble clustering analysis algorithm.Firstly,the autoencoder is used to extract key features of maneuvering trajectory to remove the impacts of redundant variables and reduce the data dimension;Then,taking the time information into account,the segmentation of Maneuver characteristic time series is realized with the improved FSTS-AEGG algorithm,and a large number of maneuver primitives are extracted;Finally,the maneuver primitives are grouped into some categories by using the selective ensemble multiple time series clustering algorithm,which can prove that each class represents a maneuver action.The maneuver pattern extraction method is applied to small scale air combat trajectory and can recognize and correctly partition at least 71.3%of maneuver actions,indicating that the method is effective and satisfies the requirements for engineering accuracy.In addition,this method can provide data support for various target maneuvering recognition methods proposed in the literature,greatly reduce the workload and improve the recognition accuracy.

Analysis of Electricity Consumption Pattern Clustering and Electricity Consumption Behavior

Studying user electricity consumption behavior is crucial for understanding their power usage patterns.However,the traditional clustering methods fail to identify emerging types of electricity consumption behavior.To address this issue,this paper introduces a statistical analysis of clusters and evaluates the set of indicators for power usage patterns.The fuzzy C-means clustering algorithm is then used to analyze 6 months of electricity consumption data in 2017 from energy storage equipment,agricultural drainage irrigation,port shore power,and electric vehicles.Finally,the proposed method is validated through experiments,where the Davies-Bouldin index and profile coefficient are calculated and compared.Experiments showed that the optimal number of clusters is 4.This study demonstrates the potential of using a fuzzy C-means clustering algorithmin identifying emerging types of electricity consumption behavior,which can help power system operators and policymakers to make informed decisions and improve energy efficiency.

Inertial Entropy and External Validation of Clusterings

Axiomatization of Shannon entropy is a subject that has received lots of attention in the information theory literature.While Shannon entropy is defined on probability distribution,we define a new type of entropy on the set of partitions of finite subsets of metric spaces,which has a rich algebraic structure as a partially ordered set.We propose an axiomatization of an entropy-like measure of partitions of sets of objects located in metric spaces,and we derive an analytic expression of this new type of entropy referred to as inertial entropy.This approach starts with the notion of inertia of a partition and includes a study of the behavior of the sum of square errors of a partition.In this context,we characterize the chain of partitions produced by the Ward hierarchical clustering method.Starting from inertial entropies of partitions,we introduce conditional entropies which,in turn,generate metrics on partitions of finite sets.These metrics are used as external validation tools for clusterings of labeled data sets.The metric generated by inertial entropy can be used to validate data clustering for labeled data sets.This type of validation aims to determine to what extend labeling of the data coincides with the clustering obtained algorithmically,and we obtain a high degree of consistency of the data labeling with the results of several hierarchical clusterings.