Good Clustering Research Articles

Learning to cluster for rendering with many lights

We present an unbiased online Monte Carlo method for rendering with many lights. Our method adapts both the hierarchical light clustering and the sampling distribution to our collected samples. Designing such a method requires us to make clustering decisions under noisy observation, and making sure that the sampling distribution adapts to our target. Our method is based on two key ideas: a coarse-to-fine clustering scheme that can find good clustering configurations even with noisy samples, and a discrete stochastic successive approximation method that starts from a prior distribution and provably converges to a target distribution. We compare to other state-of-the-art light sampling methods, and show better results both numerically and visually.

Implementation of the Simple Additive Weighting Method in Determining Centroids in the Process of Clustering the Poor in Kakatpenjalin Village, Lamongan Regency

Clustering is an algorithm in a decision support system that functions to organize an object into groups of data. In the clustering process, of course, a cluster centre is needed by the desired data group. However, the clustering process has a problem. Related research states that the results of k-means clustering can influence the selection of cluster centre points. Random selection of cluster centre points can result in different clustering results in the same data group. Not only on k-means, but k-medoids also have the same problem. So that to produce a good cluster, you must start by choosing the right cluster centre. To solve this problem, the Simple Additive Weighting method is used to select the centre point of the cluster. Simple Additive Weighting selects the centre point of the cluster by adding and summarizing the dataset. The summation is done by giving weight to each criterion and each criterion has its alternative value. From this weighted addition, the final value will be obtained. From the sum of SAW, then one of the objects with the highest and lowest values ​​can be taken to serve as the centre of the cluster.

Adversarial flow-based model for unsupervised fault diagnosis of rolling element bearings

Nowadays, numerous supervised deep learning models have been applied to bearing fault diagnosis. However, labelling the health states of the bearing vibration data is a time-consuming work and dependent on expert experience. In order to tackle this problem, a novel unsupervised bearing fault diagnosis method named adversarial flow-based model is explored in this paper. Flow-based model is a type of generative models that is proved to be better than other types in many aspects. This paper introduces the flow-based model into the field of machinery fault diagnosis, and designs an appropriate model architecture so as to train the model in unsupervised and adversarial ways. The proposed model contains an autoencoder (AE), a flow-based model, and a classifier. Firstly, the AE maps the vibration data from signal space to latent vector space. Then, the flow-based model aligns the distributions of the latent vectors of different bearing states with specific prior distributions. Finally, the classifier tries to discriminate the aligned latent vectors from the vectors sampled from the prior distributions. With the help of distinguishable prior distributions and the adversarial training mechanism between the classifier and the flow-based model together with the AE, the bearing data with the same health states are clustered into the same areas. The good clustering performance of the adversarial flow-based model is verified by a dataset with 10 health states from a bearing test rig.

MD-SPKM: A set pair k-modes clustering algorithm for incomplete categorical matrix data

In order to solve the clustering problem with incomplete and categorical matrix data sets, and considering the uncertain relationship between samples and clusters, a set pair k-modes clustering algorithm is proposed (MD-SPKM). Firstly, the correlation theory of set pair information granule is introduced into k-modes clustering. By improving the distance formula of traditional k-modes algorithm, a set pair distance measurement method between incomplete matrix samples is defined. Secondly, considering the uncertain relationship between the sample and the cluster, the definition of the intra-cluster average distance and the threshold calculation formula to determine whether the sample belongs to multiple clusters is given, and then the result of set pair clustering is formed, which includes positive region, boundary region and negative region. Finally, through the selected three data sets and four contrast algorithms for experimental evaluation, the experimental results show that the set pair k-modes clustering algorithm can effectively handle incomplete categorical matrix data sets, and has good clustering performance in Accuracy, Recall, ARI and NMI.

Design of Automatic Scoring System for Oral English Test Based on Sequence Matching and Big Data Analysis

With the application of an automatic scoring system to all kinds of oral English tests at all levels, the efficiency of test implementation has been greatly improved. The traditional speech signal processing method only focuses on the extraction of scoring features, which could not ensure the accuracy of the scoring algorithm. Aiming at the reliability of the automatic scoring system, based on the principle of sequence matching, this paper adopts the spoken speech feature extraction method to extract the features of spoken English test pronunciation and establishes a dynamic optimized spoken English pronunciation signal model based on sequence matching, which could maintain good dynamic selection and clustering ability in a strong interference environment. According to the comprehensive experiment, the automatic scoring result of the system is much higher than that of the traditional method, which greatly improves the recognition ability of oral pronunciation, solves the difference between the automatic scoring of the system and the manual scoring, and promotes the computer automatic scoring system to replace or partially replace the manual marking.

Using Partial Differential Equation Face Recognition Model to Evaluate Students’ Attention in a College Chinese Classroom

The partial differential equation learning model is applied to another high-level visual-processing problem: face recognition. A novel feature selection method based on partial differential equation learning model is proposed. The extracted features are invariant to rotation and translation and more robust to illumination changes. In the evaluation of students’ concentration in class, this paper firstly uses the face detection algorithm in face recognition technology to detect the face and intercept the expression data, and calculates the rise rate. Then, the improved model of concentration analysis and evaluation of a college Chinese class is used to recognize facial expression, and the corresponding weight is given to calculate the expression score. Finally, the head-up rate calculated at the same time is multiplied by the expression score as the final concentration score. Through the experiment and analysis of the experimental results in the actual classroom, the corresponding conclusions are drawn and teaching suggestions are provided for teachers. For each face, a large neighborhood set is firstly selected by the k -nearest neighbor method, and then, the sparse representation of sample points in the neighborhood is obtained, which effectively combines the locality of k -nearest neighbor and the robustness of sparse representation. In the sparse preserving nonnegative block alignment algorithm, a discriminant partial optimization model is constructed by using sparse reconstruction coefficients to describe local geometry and weighted distance to describe class separability. The two algorithms obtain good clustering and recognition results in various cases of real and simulated occlusion, which shows the effectiveness and robustness of the algorithm. In order to verify the reliability of the model, this paper verified the model through in-class practice tests, teachers’ questions, and interviews with students and teachers. The results show that the proposed joint evaluation method based on expression and head-up rate has high accuracy and reliability.

Hybrid clustering based on multi-criteria segmentation for higher education marketing

Market segmentation in higher education institutions is still rarely applied although it can assist in defining the right strategies and actions for the targeted market. The problem that often arises in market segmentation is how to exploit the preferences of students as customers. To overcome this, the combination of hybrid clustering method with multiple criteria will be applied to the case of the market segmentation for students in higher education institutions. The integration of geographic, demographic, psychographic, and behavioral criteria from students is used to get more insightful information about student preference. Data result of the integration will be processed using hybrid clustering using K-means and self organizing map (SOM) algorithm. The hybrid clustering conducted to get promising clustering result along with the visualization of segmentation. This study successfully produces five student segments. It received 1,386 as the Davies-Bouldin index (DBI) value and 2,752 as the quantization error (QE) value which indicates a good clustering result for market segmentation. In addition, the visualization of the clustering result can be seen in a hexagonal map.

Internal flow mechanism of cone-straight nozzle

Cone-straight nozzle has been widely used in well bore cleaning, assisting drilling in petroleum industries due to its good clustering properties. The structure including cone angle and throat length has also been studied by scholars and been optimized. However, the internal flow properties have not been investigated clearly especially the boundary layer flow. In this paper LES model is used to capture the small-scale flow state near the nozzle wall. The RNG k-epsilon model is used to validate the accuracy of the LES simulation, the simulation data shows a good agreement. Three different inlet velocities are considered in simulations. The velocity distribution, shear stress, boundary layer thickness, skin friction coefficient and Reynolds stress are analyzed, the boundary layer separation and transition are discussed. The state of flow inside nozzle is laminar with inlet velocity of 1 m/s and gradually transferred into turbulent with the increasing inlet velocity. The most severe turbulence is at the entrance of the throat section, the vortex structure appears at the entrance of converging section and dose not survive, the vortex structure appears in sequence along the throat section wall when the inlet velocity is set to 5 m/s and 10 m/s the flow properties along the conical nozzle are revealed clearly, the main flow resistance is mainly produced in throat section. All these works aim to provide theoretical support for the further processing optimization of the nozzle structure and reduce the flow resistance of nozzle.

Application of an innovative metabolomics approach to discriminate geographical origin and processing of black pepper by untargeted UHPLC-Q-Orbitrap-HRMS analysis and mid-level data fusion

An untargeted metabolomics approach based on ultra-high performance liquid chromatography coupled to high-resolution mass spectrometry (UHPLC-HRMS) fingerprinting was applied to investigate the metabolic differences of black pepper among three geographical origins (Sri Lanka, Vietnam, and Brazil) and two post-harvest processing (sterilized and non-sterilized spice). Principal component analysis (PCA) was employed to assess the overall clustering of samples, whereas supervised orthogonal partial least squares discriminant analysis (OPLS-DA) was effectively used for discrimination purposes. OPLS-DA models were fully validated (R2Y and Q2 values > 0.5) and the variable importance in projection (VIP) approach was employed to provide valuable data about differential metabolites with high discrimination potential (8 markers were putatively identified). For origin differentiation, three markers were highlighted with VIP values > 1.5 (i.e. reynosin, artabsinolide D, and tatridin B). Fatty acid derivates were the most frequent markers within the metabolites annotated for processing discrimination (e.g. 10,16-dihydroxyhexadecanoic acid and 9-hydroperoxy-10E-octadecenoic acid). Additionally, different combinations of mid-level data fusion of chromatographic-mass spectrometric techniques (UHPLC and gas chromatography coupled to HRMS) and proton nuclear magnetic resonance spectroscopy (1H NMR) were evaluated for the first time for geographical and processing discrimination of black pepper. The NMR-UHPLC-GC mid-level fused model was preferred among the tested fusion approaches since good sample clustering and no misclassification were achieved. Enhanced correct classification rate was achieved by mid-level data fusion compared with the findings obtained for one of the individual techniques (1H NMR fingerprinting) (from 92% to 100% of samples correctly classified). This study opens the path to new metabolomics approaches for black pepper authentication and quality control.

Hierarchical and K-means Clustering in the Line Drawing Data Shape Using Procrustes Analysis

One of the problems in the clustering process is that the objects under inquiry are multivariate measures containing geometrical information that requires shape clustering. Because Procrustes is a technique to obtaining the similarity measure of two shapes, it can become the solution. Therefore, this paper tried to use Procrustes as the main process in the clustering method. Several algorithms proposed for the shape clustering process using Procrustes were namely hierarchical the goodness-of-fit of Procrustes (HGoFP), k-means the goodness-of-fit of Procrustes (KMGoFP), hierarchical ordinary Procrustes analysis (HOPA), and k-means ordinary Procrustes analysis (KMOPA). Those algorithms were evaluated using Rand index, Jaccard index, F-measure, and Purity. Data used was the line drawing dataset that consisted of 180 drawings classified into six clusters. The results showed that the HGoFP, KMGoFP, HOPA and KMOPA algorithms were good enough in Rand index, F-measure, and Purity with 0.697 as a minimum value. Meanwhile, the good clustering results in the Jaccard index were only the HGoFP, KMGoFP, and HOPA algorithms with 0.561 as a minimum value. KMGoFP has the worst result in the Jaccard index that is about 0.300. In the time complexity, the fastest algorithm is the HGoFP algorithm; the time complexity is 4.733. Based on the results, the algorithms proposed in this paper particularly deserve to be proposed as new algorithms to cluster the objects in the line drawing dataset. Then, the HGoFP is suggested clustering the objects in the dataset used.

Identifying New Potential Biomarkers in Adrenocortical Tumors Based on mRNA Expression Data Using Machine Learning.

Simple SummaryUsing a visual-based clustering method on the TCGA RNA sequencing data of a large adrenocortical carcinoma (ACC) cohort, we were able to classify these tumors in two distinct clusters largely overlapping with previously identified ones. As previously shown, the identified clusters also correlated with patient survival. Applying the visual clustering method to a second dataset also including benign adrenocortical samples additionally revealed that one of the ACC clusters is more closely located to the benign samples, providing a possible explanation for the better survival of this ACC cluster. Furthermore, the subsequent use of machine learning identified new possible biomarker genes with prognostic potential for this rare disease, that are significantly differentially expressed in the different survival clusters and should be further evaluated.Adrenocortical carcinoma (ACC) is a rare disease, associated with poor survival. Several “multiple-omics” studies characterizing ACC on a molecular level identified two different clusters correlating with patient survival (C1A and C1B). We here used the publicly available transcriptome data from the TCGA-ACC dataset (n = 79), applying machine learning (ML) methods to classify the ACC based on expression pattern in an unbiased manner. UMAP (uniform manifold approximation and projection)-based clustering resulted in two distinct groups, ACC-UMAP1 and ACC-UMAP2, that largely overlap with clusters C1B and C1A, respectively. However, subsequent use of random-forest-based learning revealed a set of new possible marker genes showing significant differential expression in the described clusters (e.g., SOAT1, EIF2A1). For validation purposes, we used a secondary dataset based on a previous study from our group, consisting of 4 normal adrenal glands and 52 benign and 7 malignant tumor samples. The results largely confirmed those obtained for the TCGA-ACC cohort. In addition, the ENSAT dataset showed a correlation between benign adrenocortical tumors and the good prognosis ACC cluster ACC-UMAP1/C1B. In conclusion, the use of ML approaches re-identified and redefined known prognostic ACC subgroups. On the other hand, the subsequent use of random-forest-based learning identified new possible prognostic marker genes for ACC.

Fast spectral clustering method based on graph similarity matrix completion

Spectral clustering (SC) is a widely used technique to perform group unsupervised classification of graph signals. However, SC is sometimes computationally intensive due to the need to calculate the graph similarity matrices on large high-dimensional data sets. This paper proposes an efficient SC method that rapidly calculates the similarity matrix using a matrix completion algorithm. First, a portion of the elements in the similarity matrix are selected by a blue noise sampling mask, and their similarity values are calculated directly from the original dataset. After that, a split Bregman algorithm based on the Schatten capped p norm is developed to rapidly retrieve the rest of the matrix elements. Finally, spectral clustering is performed based on the completed similarity matrix. A set of simulations based on different data sets are used to assess the performance of the proposed method. It is shown that for a sufficiently large sampling rate, the proposed method can accurately calculate the completed similarity matrix, and attain good clustering results while improving on computational efficiency.

Algorithms for fair k-clustering with multiple protected attributes

We study fair center based clustering problems. In an influential paper, Chierichetti, Kumar, Lattanzi and Vassilvitskii (NIPS 2017) consider the problem of finding a good clustering, say of women and men, such that every cluster contains an equal number of women and men. They were able to obtain a constant factor approximation for this problem for most center based k-clustering objectives such as k-median, k-means, and k-center. Despite considerable interest in extending this problem for multiple protected attributes (e.g. women and men, with or without citizenship), so far constant factor approximations for these problems have remained elusive except in special cases. We settle this question in the affirmative by giving the first constant factor approximation for a wide range of center based k-clustering objectives.

The utility of clusters and a Hungarian clustering algorithm.

Implicit in the k–means algorithm is a way to assign a value, or utility, to a cluster of points. It works by taking the centroid of the points and the value of the cluster is the sum of distances from the centroid to each point in the cluster. The aim in this paper is to introduce an alternative way to assign a value to a cluster. Motivation is provided. Moreover, whereas the k–means algorithm does not have a natural way to determine k if it is unknown, we can use our method of evaluating a cluster to find good clusters in a sequential manner. The idea uses optimizations over permutations and clusters are set by the cyclic groups; generated by the Hungarian algorithm.

An empirical comparison between stochastic and deterministic centroid initialisation for K-means variations

K-Means is one of the most used algorithms for data clustering and the usual clustering method for benchmarking. Despite its wide application it is well-known that it suffers from a series of disadvantages; it is only able to find local minima and the positions of the initial clustering centres (centroids) can greatly affect the clustering solution. Over the years many K-Means variations and initialisation techniques have been proposed with different degrees of complexity. In this study we focus on common K-Means variations along with a range of deterministic and stochastic initialisation techniques. We show that, on average, more sophisticated initialisation techniques alleviate the need for complex clustering methods. Furthermore, deterministic methods perform better than stochastic methods. However, there is a trade-off: less sophisticated stochastic methods, executed multiple times, can result in better clustering. Factoring in execution time, deterministic methods can be competitive and result in a good clustering solution. These conclusions are obtained through extensive benchmarking using a range of synthetic model generators and real-world data sets.

Multi-stage perforation and hydraulic fracture stage selection based on machine learning methods

In unconventional oil and gas resources, especially shale oil and gas resources with extremely low permeability and porosity, in order to develop effectively, it is necessary to establish perforation and multi-stage fracturing. The machine learning algorithm k-means clustering was used to cluster the five features selected to describe reservoir properties: shale content, porosity, total organic carbon content and those reflecting rock mechanical properties: Young’s modulus and Poisson’s ratio. The data were classified in a high-dimensional space to determine different perforation fracture stages. The classification algorithm XGboost was then used to predict different perforation stages using four conventional logging curves GR, NPRL, VP and DEN. The K-means clustering algorithm based on Euclidean distance can well classify the selected features in the high-dimensional space. The Hopkins statistics of clustering trend is 0.94, showing a good clustering trend. When the classification algorithm is used for prediction, the average accuracy is 0.92, the average recall rate is 0.90, and the average F1 score is 0.90, which can predict different perforated fracture stages well and optimize the design of the perforated fracture stage.

A self-adaptive DBSCAN-based method for wafer bin map defect pattern classification

The wafer map is obtained by testing each die in the wafer during semiconductor production for defects and marking the defective die. The classification of wafer maps can provide evidence for problems occurring in the production process, so as to solve the problems and reduce the costs. Before classifying the wafer map, the most important thing is feature extraction. In addition to a certain spatial pattern, the wafer map also has a lot of noise, which affects the process of feature extraction. When the traditional DBSCAN algorithm is used for filtering, it is necessary to manually determine the value of Eps and MinPts parameters, and the selection of the parameters directly affects the accuracy of the clustering. Therefore, this paper proposes an automatic parameter filtering method based on DBSCAN, which can solve the traditional drawbacks of manually parameters setting, the algorithm is a Self-Adaptive DBSCAN-based method for wafer bin map, we call it SA-DBSCANWBM. This method selects a comprehensive index of cluster intra-cluster density and inter-cluster density to evaluate the optimal parameters. The experimental results show that the algorithm proposed in this paper can automatically and reasonably select better parameters and has a good clustering effect, which helpful for subsequent feature extraction and classification.

A Controllable RSU and Vampire Moth to Support the Cluster Stability in VANET

VANET has a dynamic topology that affects cluster formation stability. It influences vehicle’s network quality though supporting this stability requires a fast and small cluster formation process. It is necessary because of the rapidly changing condition of the vehicle's position. Moreover, small cluster groups make network quality more evenly distributed among its members. They are essential components in the formation of good clusters in VANET. In the previous research, the algorithm used for this process is based on moth flame optimization. This study proposes modifications in this algorithm to speed up the convergence process, facilitated by a model of turning angles in the moth. Furthermore, the flame at the destination moth is created dynamically to approach the real conditions in VANET. The use of coefficients is further introduced to reduce cluster size according to the traffic conditions. The cluster management process uses a fully controlled Road Side Unit on the vehicle’s traffic conditions. In the final result, the convergence time is slightly faster, with better results in the throughput at 101.1%, the number of clusters at 58.1%, and the delay at 5.5%, respectively.

Ship AIS Trajectory Clustering: An HDBSCAN-Based Approach

The Automatic Identification System (AIS) of ships provides massive data for maritime transportation management and related researches. Trajectory clustering has been widely used in recent years as a fundamental method of maritime traffic analysis to provide insightful knowledge for traffic management and operation optimization, etc. This paper proposes a ship AIS trajectory clustering method based on Hausdorff distance and Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN), which can adaptively cluster ship trajectories with their shape characteristics and has good clustering scalability. On this basis, a re-clustering method is proposed and comprehensive clustering performance metrics are introduced to optimize the clustering results. The AIS data of the estuary waters of the Yangtze River in China has been utilized to conduct a case study and compare the results with three popular clustering methods. Experimental results prove that this method has good clustering results on ship trajectories in complex waters.

A fuzzy spectral clustering algorithm for hyperspectral image classification

Spectral clustering is an unsupervised clustering algorithm, and is widely used in the field of pattern recognition and computer vision due to its good clustering performance. However, the traditional spectral clustering algorithm is not suitable for large-scale data classification, such as hyperspectral remote sensing image, because of its high computational complexity, and it is difficult to characterize the inherent uncertainty of the hyperspectral remote sensing image. This paper uses fuzzy anchors to process hyperspectral image classification and proposes a novel spectral clustering algorithm based on fuzzy similarity measure. The proposed algorithm utilizes the fuzzy similarity measure to obtain the similarity between the data points and the anchors, and then gets the similarity matrix. Finally, spectral clustering is performed on the similarity matrix to compute the classification results. The experimental results on the hyperspectral remote sensing image data sets have demonstrated the effectiveness of the proposed algorithm, and the introduction of fuzzy similarity measure gives rise to a more robust similarity matrix. Compared with existing methods, the proposed algorithm has a better classification result on the hyperspectral remote sensing image, and the kappa coefficient obtained by the proposed algorithm is 2% higher than the traditional algorithms.