Entropy Clustering Research Articles

Enhancing multi-label disease diagnosis through hypergraph clustering and multi-classification label entropy

Enhancing multi-label disease diagnosis through hypergraph clustering and multi-classification label entropy

Reducing the Impact of Time Evolution on Source Code Authorship Attribution via Domain Adaptation

Source code authorship attribution is an important problem in practical applications such as plagiarism detection, software forensics, and copyright disputes. Recent studies show that existing methods for source code authorship attribution can be significantly affected by time evolution, leading to a decrease in attribution accuracy year by year. To alleviate the problem of Deep Learning (DL)-based source code authorship attribution degrading in accuracy due to time evolution, we propose a new framework called Time D omain A daptation (TimeDA) by adding new feature extractors to the original DL-based code attribution framework that enhances the learning ability of the original model on source domain features without requiring new or more source data. Moreover, we employ a centroid-based pseudo-labeling strategy using neighborhood clustering entropy for adaptive learning to improve the robustness of DL-based code authorship attribution. Experimental results show that TimeDA can significantly enhance the robustness of DL-based source code authorship attribution to time evolution, with an average improvement of 8.7% on the Java dataset and 5.2% on the C++ dataset. In addition, our TimeDA benefits from employing the centroid-based pseudo-labeling strategy, which significantly reduced the model training time by 87.3% compared to traditional unsupervised domain adaptive methods.

Brain image segmentation with fuzzy entropy clustering and PSO-GWO optimization techniques

In the field of brain MRI analysis, image segmentation serves various purposes such as quantifying and visualizing anatomical structures, analyzing brain changes, delineating pathological regions, and aiding in surgical planning and image-guided interventions. Over the past few decades, diverse segmentation techniques with varying degrees of accuracy and complexity have been developed. Real-world brain MRI images often encounter intensity in homogeneity, posing a significant challenge in accurate segmentation. The prevailing image segmentation algorithms, predominantly region-based, typically rely on the homogeneity of image intensities in specific regions of interest. However, these methods often fall short of providing precise segmentation results due to intensity in homogeneity. To address these challenges and enhance segmentation performance, this paper introduce a novel objective function named Fuzzy Entropy Clustering with Local Spatial Information and Bias Correction (FECSB). Additionally, we propose a novel hybrid algorithm that combines Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO) to maximize the effectiveness of the FECSB function in MRI brain image segmentation. The proposed algorithm undergoes rigorous evaluation using benchmark MRI brain images, including those from the McConnell Brain Imaging Center (BrainWeb). The experimental results unequivocally demonstrate the superiority of the PSO-GWO clustering method over the traditional Fuzzy C Means (FCM) method. Across various image slices, the PSO-GWO method consistently outperforms FCM in terms of accuracy, showing improvements ranging from 1.28% to 1.46%, approximately achieving 99.37% accuracy.

Similarity analysis study of the carbon emission metrics through fuzzy entropy clustering

The world has changed to a massive degree in the past thousands of years. Most of the time, the amount of carbon dioxide in the atmosphere remains constant. In the late 18th century, according to the sources of CDIAC and NOOA, the level of carbon dioxide began to rise, and then in the 20th century, it went through the roof, reaching levels that had not been seen in nature for millions of years. The increase in carbon in the atmosphere is the major contributing factor to climate change. The key to reversing the damage is restoring the earth’s delicate, balanced carbon cycle. As carbon cycle depicts the way carbon moves around the earth. It consists of sources that emit the carbon component into the atmosphere. The biological side of the carbon cycle is well balanced due to respiration, where carbon dioxide is released into the atmosphere, then plants, bacteria, and algae take carbon dioxide out of the atmosphere during photosynthesis and the process they use to generate chemical energy. On the other hand, oceans are the best sources and sinks; carbon dioxide is endlessly being absorbed into the ocean and released from the oceans almost exactly at the same rate, which is rapidly influencing the carbon cycle. Similarity is a methodology that has many applications in the real world. The current research article is destined to study how statistics of carbon emission metrics are alike and belong to one cluster. In the current study, the research is destined to derive a similarity analysis of several countries’ carbon emission metrics that are alike and often fall in the range of [0, 1]. And deriving the proximity of the carbon emission metrics leading to similarity or dissimilarity. In the current context of data matrices of numerical data, an Euclidian measure of distance between two data elements will yield a degree of similarity. The current research article is destined to study the similarity analysis of carbon emission metrics through fuzzy entropy clustering.

Multi-scale reconstruction of rolling mill vibration signal based on fuzzy entropy clustering

The mill vibration signals collected during mill operation are often affected by complex working conditions and various disturbances, thus presenting strong non-stationary characteristics. After the vibration signal is processed by noise reduction, although the noise interference and modal aliasing phenomenon are solved, the extracted vibration signal of the rolling mill still has the problems of strong non-stationarity, high complexity, and great difficulty in analyzing, which largely affects the prediction model construction. To address the above problems, this study proposes a multi-scale signal reconstruction method based on Fuzzy Entropy and Gath-Geva fuzzy clustering algorithm, which effectively reduces the complexity of the original sequence, reduces the number of predictive modellings, and improves the signal predictability and prediction accuracy. The sequence reconstruction of such non-stationary vibration signals based on their internal characteristics has obvious advantages for the problem of mill vibration prediction and analysis.

A novel evaluation approach for the ability of older adults based on grey clustering model

PurposeThe purpose of this study is to advance a novel evaluation index system and evaluation approach for ability of older adults in China.Design/methodology/approachThis study constructed a comprehensive older adult ability evaluation index system with 4 primary indicators and 17 secondary indicators. Grey clustering analysis and entropy weight method are combined into a robust evaluation model for the ability of older adults.FindingsThe result demonstrates that the proposed grey clustering model is readily available to calculate the disability level of elderly individuals. The constructed index system more comprehensively considers all aspects of the disability of the elderly.Originality/valueThis study provides a quantitative method and a more reasonable index system for the determination of the disability level of the elderly.

INNOVATIONS IN THE FIELD OF SPORTS INDUSTRY MANAGEMENT: ASSESSMENT OF THE DIGITAL ECONOMY'S IMPACT ON THE QUALITATIVE DEVELOPMENT OF THE SPORTS INDUSTRY

The rapid development of the digital economy has promoted industrial transformation and upgrading, brought changes to the production mode and personal consumption mode of enterprises, and the development of traditional industries are constantly being realised in a digital format. The research of current scholars mainly focuses on: the calculation of digital economy comprehensive index, it is only scientific interpretation, lack of empirical research; second, the lack of existing research on digital economy indicators, mostly only from a single perspective; and finally, the lack of analysis of regional heterogeneity of digital economy development. Methodology. In the index system weighting method is mainly divided into subjective empowerment and objective empowerment, in terms of subjective empowerment, it is based on the relative importance between the indicators by subjective judgment to give appropriate weight, such as principal component analysis, AHP; and objective empowerment is based on the original information to empower, such as cluster analysis, standard deviation, entropy and extreme method, and so forth. In the selection of measurement method, some scholars believe that the subjective weight method will be disturbed by human factors, which will bias the calculation results, and can not respond to authenticity. In view of this, this paper adopts the entropy method to measure the high-quality economic development level and digital economy development level in 31 provinces in China from 2010 to 2019. The research of current scholars mainly focuses on: the calculation of digital economy comprehensive index, it is only scientific interpretation, lack of empirical research; second, the lack of existing research on digital economy indicators, mostly only from a single perspective; and finally, the lack of analysis of regional heterogeneity of digital economy development. This paper constructs three subsystems and 11 basic indicators to construct the evaluation index of digital economy development, and three subsystems and 9 basic indicators to evaluate the high-quality development of sports industry. To explore the spatial spillover effect of digital economy development on the high-quality development of sports industry, the high-quality development of sports industry in China in 2011-2020 was taken as the explanatory variable, and the development of digital economy was taken as the explanatory variable. The purpose of the article is to build spatial measurement model three subsystems, which include 11 basic indicators to construct the evaluation index of digital economy development, and three subsystems and 9 basic indicators to evaluate the high-quality development of sports industry. The base of research is 31 provinces, municipalities, autonomous regions of China. Conclusion. First, the digital economy is innovation as the core driving force of the new economic form, and the deep fusion of sports industry can give new impetus to the development of high-quality sports industry, further promote the development of the quality of sports industry. Second, digital economy development through advanced digital technology can directly and indirectly promote the development of sports industry, digital economy and sports industry quality development have significant spatial correlation, "high-high", "low-low" agglomeration characteristics, digital economy development can significantly promote the regional sports industry development, and the quality of sports industry development is also obvious.

A Semisupervised Soft-Sensor of Just-in-Time Learning With Structure Entropy Clustering and Applications for Industrial Processes Monitoring

To monitor industrial processes properly, soft-sensors are widely used to predict significant but difficult-to-measure quality variables. However, the prediction performances of traditional data-driven soft-sensors are usually unacceptable once suffering from high-nonlinear, high-dimension and imblance data issues. Therefore, a semi-supervised soft-sensor, which is learned by a just-in-time method with structure entropy clustering (SS-JITL-SEC), is proposed aiming to improve prediction performance with a simpler way. Inspired by a divide and conquer strategy, a novel SEC method is proposed to achieve several clusters and then to translate the highly complex and nonlinear modeling problems into simple and linear ones. Moreover, the training dataset is extended through a mixed semi-supervised (SS) labeling approach. Finally, dissimilarity-based just-in-time learning (JITL) works together with the resulting clustering sub-datasets to formulate a local adaptive prediction model. Two datasets from different types of wastewater treatment plants are used to verify the effectiveness of the proposed soft-sensor. The results show that the SS-JITL-SEC soft-sensor can achieve better prediction performance than other standard counterparts, and even for effective process monitoring with the resulted residuals.

Risk assessment and configuration of water and land resources system network in the Huang-Huai-Hai watershed

Under the influence of global warming and human activities, the risk sources of land and water resources system networks are complex and uncertain, which has induced a series of ecological and environmental problems. It is of great significance to systematically identify and evaluate the risks of water and land resources system networks from the perspective of water resource attributes and land resource attributes. Therefore, based on the mutual feedback mechanism of water and land resources, this study constructs the water and land resources system network in the Huang-Huai-Hai Watershed. The cluster analysis-ISODATA algorithm and entropy weight-TOPSIS evaluation method were used to evaluate and predict the risk of the water and land resources system network in the Huang-Huai-Hai Watershed. In addition, a model for the optimal allocation of water and land resources for risk mitigation was constructed, and the optimal allocation scheme of water and land resources in the Huang-Huai-Hai Watershed in 2030, 2035 and 2060 under different scenario modes was obtained. The results show that the risk of the water and land resources system network in the Huang-Huai-Hai Watershed from 1990 to 2020 first increased and then decreased, and the proportion of areas above medium–high risk in 2001–2010 was 53.17%. Compared with 2020, the area above medium–high risk in 2030, 2035 and 2060 increased by 1.4–17.38%, 6.78–15.92% and 2.39–6.82%, respectively. After the optimal allocation of water and land resources for risk mitigation, the targets of optimal allocation of water and land resources in 2030, 2035 and 2060 were improved, among which the water shortage under different scenario modes was reduced by 71.92–79.19%, 65.92–74.46% and 62.23–71.69%, and the area above medium–high risk in the water and land resources system network decreased by 48.4–59.79%, 20.99–59.7% and 29.53–49.06%, respectively. The results of this study can provide technical support for coping with global climate change, solving the problem of water shortage and optimizing the spatial layout of the land in the Huang-Huai-Hai Watershed.

Functional brain network features specify DBS outcome for patients with treatment resistant depression.

Deep brain stimulation (DBS) has shown therapeutic benefits for treatment resistant depression (TRD). Stimulation of the subcallosal cingulate gyrus (SCG) aims to alter dysregulation between subcortical and cortex. However, the 50% response rates for SCG-DBS indicates that selection of appropriate patients is challenging. Since stimulation influences large-scale network function, we hypothesized that network features can be used as biomarkers to inform outcome. In this pilot project, we used resting-state EEG recorded longitudinally from 10 TRD patients with SCG-DBS (11 at baseline). EEGs were recorded before DBS-surgery, 1-3 months, and 6 months post surgery. We used graph theoretical analysis to calculate clustering coefficient, global efficiency, eigenvector centrality, energy, and entropy of source-localized EEG networks to determine their topological/dynamical features. Patients were classified as responders based on achieving a 50% or greater reduction in Hamilton Depression (HAM-D) scores from baseline to 12 months post surgery. In the delta band, false discovery rate analysis revealed that global brain network features (segregation, integration, synchronization, and complexity) were significantly lower and centrality of subgenual anterior cingulate cortex (ACC) was higher in responders than in non-responders. Accordingly, longitudinal analysis showed SCG-DBS increased global network features and decreased centrality of subgenual ACC. Similarly, a clustering method separated two groups by network features and significant correlations were identified longitudinally between network changes and depression symptoms. Despite recent speculation that certain subtypes of TRD are more likely to respond to DBS, in the SCG it seems that underlying brain network features are associated with ability to respond to DBS. SCG-DBS increased segregation, integration, and synchronizability of brain networks, suggesting that information processing became faster and more efficient, in those patients in whom it was lower at baseline. Centrality results suggest these changes may occur via altered connectivity in specific brain regions especially ACC. We highlight potential mechanisms of therapeutic effect for SCG-DBS.

Robust stabilization for a class of complex production process systems based on multi-dimensional moving pattern

This paper studies robust stabilization problem and state feedback controller design problem for a class of complex production process systems by pattern class variable instead of state variable or output variable. Firstly, the maximum entropy clustering method is introduced and modeling method based on multi-dimensional moving pattern is given. Nonlinear state space models of systems with input-delay are established by using the trajectories of pattern class variable. Then sufficient conditions for robust stabilization of the system are obtained by Linear Matrix Inequality. In addition, a state feedback controller is designed based on this condition. Finally, the effectiveness of the methods is illustrated by taking the actual working conditions data as an example.

An elementary-step kinetic model for ethyl acetate synthesis by direct addition of acetic acid to ethylene on a silicotungstic acid catalyst

An elementary-step kinetic model for ethyl acetate synthesis by direct addition of acetic acid to ethylene over a silica supported silicotungstic acid catalyst is developed. A reaction network comprising 5 reactions, i.e. 3 direct additions, an esterification and a condensation, between 6 major components (ethylene, acetic acid, water, ethyl acetate, ethanol and diethyl ether) is accounted for. The reactions are described by means of Langmuir-Hinshelwood-Hougen-Watson and Eley-Rideal type rate expressions, including the adsorptions of all components, except diethyl ether, on the active sites of the catalyst. The model can quantitatively capture the effects of space time, temperature, pressure and feed composition, on conversions and also qualitatively reproduces the observed selectivities. However, it is necessary to include two physicochemical phenomena, i.e. i) silicotungstic acid solvation by water for the adequate description of the pressure effect on kinetics and ii) the proton-ethanol cluster formation to improve the fit between the experimental and model predicted values. Estimates for activation energies, adsorption enthalpies, solvation and cluster formation enthalpies and entropies are obtained through regression. At a temperature of 442 K, pressure equal to 1.2 MPa, an ethylene:acetic acid:water:dinitrogen feed molar ratio of 78.2:6.5:5.3:10.0 and relative space time amounting to 0.5, ethyl acetate is for 84 % produced by direct addition of acetic acid to ethylene, while the esterification between acetic acid and ethanol contributes for 16 %. Hence, besides reproducing conversions and selectivities, the kinetic model provides insights on fundamental level, which are essential for the optimization and innovation of production plants utilizing direct addition technology.

Study on numerical simulation method of viscosity time-varying slurry diffusion law

In this study, finite element and finite difference methods were used for numerical calculations. The law of slurry diffusion in the inclined plane cracks is summarized in this study. Further, the equation describing the diffusion of slurry with time-varying viscosity in the plane fracture with dip angle was derived, and the mathematical model of slurry diffusion control equation and improved entropy clustering algorithm were developed. Thus, the law of slurry diffusion with time-varying viscosity in an inclined plane fractured rock mass was studied in detail, and the control equation obtained was modified to improve the prediction accuracy of diffusion of slurry with time-varying viscosity. By applying the model to actual working conditions, the results highlighted that the improved entropy clustering algorithm could more accurately describe the relationship between the factors influencing slurry diffusion and the slurry diffusion radius. The improved entropy clustering algorithm model can adjust the weights of the factors influencing the slurry diffusion in the diffusion control equation, which leads to a more flexible and more accurate slurry diffusion control equation.

ADDPC-SMOTE: An Oversampling Algorithm Based on Density Difference Peak Clustering and Spatial Distribution Entropy

ADDPC-SMOTE: An Oversampling Algorithm Based on Density Difference Peak Clustering and Spatial Distribution Entropy

Study on the Drug Use Pattern of Professor Li Jun's Treatment of Coronary Heart Disease Based on Association Rules and Entropy Clustering of Complex Systems

Study on the Drug Use Pattern of Professor Li Jun's Treatment of Coronary Heart Disease Based on Association Rules and Entropy Clustering of Complex Systems

System asymptotic stability analysis of a kind of complex production processes based on multi-dimensional moving pattern

This paper presents a new method to study the system stability of a kind of complex production processes. The system dynamics is described by multi-dimensional pattern class variables instead of state variables or output variables. The system asymptotic stability is defined by system matrix and output class partition. Firstly, “pattern moving space” is constructed by a new maximum entropy clustering method. Multi-dimensional pattern class variables measured by pattern class centers are defined in this new space. Then the nonlinear state space model of the relevant control system is derived and established through two-step prediction model. On the basis of the original stability in Euclidean space, the multi-dimensional pattern moving state space is defined and Lyapunov stability is given in this space. Finally, experimental results show the effectiveness of the proposed conditions.

Information in Streetscapes—Research on Visual Perception Information Quantity of Street Space Based on Information Entropy and Machine Learning

Urban street space is a critical reflection of a city’s vitality and image and a critical component of urban planning. While visual perceptual information about an urban street space can reflect the composition of place elements and spatial relationships, it lacks a unified and comprehensive quantification system. It is frequently presented in the form of element proportions without accounting for realistic factors, such as occlusion, light and shadow, and materials, making it difficult for the data to accurately describe the complex information found in real scenes. The conclusions of related studies are insufficiently focused to serve as a guide for designing solutions, remaining merely theoretical paradigms. As such, this study employed semantic segmentation and information entropy models to generate four visual perceptual information quantity (VPIQ) measures of street space: (1) form; (2) line; (3) texture; and (4) color. Then, at the macro level, the streetscape coefficient of variation (SCV) and K-means cluster entropy (HCK) were proposed to quantify the street’s spatial variation characteristics based on VPIQ. Additionally, we used geographically weighted regression (GWR) to investigate the relationship between VPIQ and street elements at the meso level as well as its practical application. This method can accurately and objectively describe and detect the current state of street spaces, assisting urban planners and decision-makers in making decisions about planning policies, urban regeneration schemes, and how to manage the street environment.

A Hybrid Model for China’s Soybean Spot Price Prediction by Integrating CEEMDAN with Fuzzy Entropy Clustering and CNN-GRU-Attention

China’s soybean spot price has historically been highly volatile due to the combined effects of long-term massive import dependence and intricate policies, as well as inherent environmental elements. The accurate prediction of the price is crucial for reducing the amount of soybean-linked risks worldwide and valuable for the long-term sustainability of global agriculture. Therefore, a hybrid prediction model that combines component clustering and a neural network with an attention mechanism has been developed. After fully integrated complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) processing of the price series, the fuzzy entropy of each component is measured as the complexity characteristic. K-means clustering and reconstruction are applied to the components before being input to the CNN-GRU-Attention network for prediction to improve the model ability and adaptability of the sequences. In the empirical analysis, the proposed model outperforms other decomposition techniques and machine learning algorithms regarding prediction accuracy. After applying the decomposition part, the results have RMSE, MAPE, and MAE values of 49.59%, 22.58%, and 21.99% lower than those of the individual prediction part, respectively. This research presents a novel approach for market participants in the soybean industry for risk response. It gives a new perspective on agricultural product prices in sustainable agricultural marketing, while also providing practical tools for developing public policies and decision-making.

A statistical mechanics framework for immiscible and incompressible two-phase flow in porous media

We construct a statistical mechanics for immiscible and incompressible two-phase flow in porous media under local steady-state conditions based on the Jaynes maximum entropy principle. A cluster entropy is assigned to our lack of knowledge of, and control over, the fluid and flow configurations in the pore space. As a consequence, two new variables describing the flow emerge: The agiture, which describes the level of agitation of the two fluids, and the flow derivative, which is conjugate to the saturation. Agiture and flow derivative are the analogs of temperature and chemical potential in standard (thermal) statistical mechanics. The associated thermodynamics-like formalism reveals a number of hitherto unknown relations between the variables that describe the flow, including fluctuations. The formalism opens for new approaches to characterize porous media with respect to multi-phase flow for practical applications, replacing the simplistic relative permeability theory while still keeping the number of variables tractable.

Relative cluster entropy for power-law correlated sequences

We propose an information-theoretical measure, the relative cluster entropy \mathcal{D_C}[P\|Q]𝒟𝒞[P∥Q], to discriminate among cluster partitions characterised by probability distribution functions P and Q. The measure is illustrated with the clusters generated by pairs of fractional Brownian motions with Hurst exponents H_1H1 and H_2H2 respectively. For subdiffusive, normal and superdiffusive sequences, the relative entropy sensibly depends on the difference between H_1H1 and H_2H2. By using the minimum relative entropy principle, cluster sequences characterized by different correlation degrees are distinguished and the optimal Hurst exponent is selected. As a case study, real-world cluster partitions of market price series are compared to those obtained from fully uncorrelated sequences (simple Browniam motions) assumed as a model. The minimum relative cluster entropy yields optimal Hurst exponents H_1H1=0.55, H_1H1=0.57, and H_1H1=0.63 respectively for the prices of DJIA, S and P500, NASDAQ: a clear indication of non-markovianity. Finally, we derive the analytical expression of the relative cluster entropy and the outcomes are discussed for arbitrary pairs of power-laws probability distribution functions of continuous random variables.