Constrained K-means Algorithm Research Articles

Balanced clustering for student admission school zoning by parameter tuning of constrained k-means

The Indonesian government issued a regulation through the Ministry of Education and Culture, number 51 of 2018, which contains zoning rules to improve the quality of education in school educational institutions. This research aims to compare the performance of the k-means algorithm with the constrained k-means algorithm to model the zoning of each school area based on the shortest distance parameter between the school location and the domicile of prospective students. The study used data from 2248 prospective students and 22 public school locations. The results of testing the k-means algorithm in grouping showed the formation of non-circular patterns in the cluster membership with different numbers of centroid cluster members. In contrast, testing the constrained k-means algorithm showed balanced outcomes in cluster membership with a membership value of 103 for each school as the cluster center. The research findings state that the developed constrained k-means algorithm solves the problem of unbalanced data clustering and overlapping issues in the process of new student admissions. In other words, the constrained k-means algorithm can be a reference for the government in making decisions on new student admissions

Constrained k-means algorithm for resource allocation in mobile cloudlets

Constrained k-means algorithm for resource allocation in mobile cloudlets

Improved Constrained k-Means Algorithm for Clustering with Domain Knowledge

Witnessing the tremendous development of machine learning technology, emerging machine learning applications impose challenges of using domain knowledge to improve the accuracy of clustering provided that clustering suffers a compromising accuracy rate despite its advantage of fast procession. In this paper, we model domain knowledge (i.e., background knowledge or side information), respecting some applications as must-link and cannot-link sets, for the sake of collaborating with k-means for better accuracy. We first propose an algorithm for constrained k-means, considering only must-links. The key idea is to consider a set of data points constrained by the must-links as a single data point with a weight equal to the weight sum of the constrained points. Then, for clustering the data points set with cannot-link, we employ minimum-weight matching to assign the data points to the existing clusters. At last, we carried out a numerical simulation to evaluate the proposed algorithms against the UCI datasets, demonstrating that our method outperforms the previous algorithms for constrained k-means as well as the traditional k-means regarding the clustering accuracy rate although with a slightly compromised practical runtime.

Clustering and Dispatching Rule Selection Framework for Batch Scheduling

In this study, a batch scheduling with job grouping and batch sequencing is considered. A clustering algorithm and dispatching rule selection model is developed to minimize total tardiness. The model and algorithm are based on the constrained k-means algorithm and neural network. We also develop a method to generate a training dataset from historical data to train the neural network. We use numerical examples to demonstrate that the proposed algorithm and model efficiently and effectively solve batch scheduling problems.

A bag of constrained informative deep visual words for image retrieval

In this paper, we propose a bag of constrained informative deep visual words (BoCIDVW) model for image retrieval. Informative patches from each image are first obtained using patch entropy values. Each such patch is represented by deep features extracted through VGG16-Net. Two sets of constraints, namely, the must-link (ML) and the cannot-link (CL), are obtained for each deep informative patch in an unsupervised manner from its mutual information values (with other patches). The patches are then quantized using the Linear-time Constrained Vector Quantization Error (LCVQE), a fast yet accurate constrained K-means algorithm. The resulting clusters, which we term constrained informative deep visual words, are employed to label each patch. Finally, a bag (histogram) of constrained informative visual words is developed for image retrieval. Experiments on three different publicly available datasets demonstrate the merit of the proposed formulation.

Boosted Constrained K-Means Algorithm for Social Networks Circles Analysis

The volume of information generated by a huge number of social networks users is increasing every day. Social networks analysis has gained intensive attention in the data mining research community to identify circles of users depending on the characteristics in the individual profiles or the structure of the network. In this paper, we propose the boosting principle to find the circles of social networks. Constrained k-means clustering method is used as a weak learner with the boosting framework. This method generates a constrained clustering represented by a kernel matrix according to the priorities of the pair-wise constraints. The experimental results show that the proposed algorithm using boosting principle for social network analysis improves the performance of the clustering and outperforms the state-of-the-art.

Clustering Using Boosted Constrained k-Means Algorithm

This article proposes a constrained clustering algorithm with competitive performance and less computation time to the state-of-the-art methods, which consists of a constrained k-means algorithm enhanced by the boosting principle. Constrained k-means clustering using constraints as background knowledge, although easy to implement and quick, has insufficient performance compared with metric learning-based methods. Since it simply adds a function into the data assignment process of the k-means algorithm to check for constraint violations, it often exploits only a small number of constraints. Metric learning-based methods, which exploit constraints to create a new metric for data similarity, have shown promising results although the methods proposed so far are often slow depending on the amount of data or number of feature dimensions. We present a method that exploits the advantages of the constrained k-means and metric learning approaches. It incorporates a mechanism for accepting constraint priorities and a metric learning framework based on the boosting principle into a constrained k-means algorithm. In the framework, a metric is learned in the form of a kernel matrix that integrates weak cluster hypotheses produced by the constrained k-means algorithm, which works as a weak learner under the boosting principle. Experimental results for 12 data sets from 3 data sources demonstrated that our method has performance competitive to those of state-of-the-art constrained clustering methods for most data sets and that it takes much less computation time. Experimental evaluation demonstrated the effectiveness of controlling the constraint priorities by using the boosting principle and that our constrained k-means algorithm functions correctly as a weak learner of boosting.

A Constrained K-Means and Nearest Neighbor Approach for Route Optimization in the Bale Collection Problem

The bale collection problem (BCP) appears after harvest operations of agricultural crops. The solution to the BCP is defined by the sequence in which bales, that lie scattered across the field, are collected. This paper presents a constrained k-means algorithm and nearest neighbor approach to the BCP, which minimizes travel time and hence fuel consumption. The constraints imposed on the k-means procedure is not the usual condition that certain groups of objects must be clustered together, but rather that the cluster centers must lie on valid locations, which may be specified as functions or sets of points. The algorithm can be incorporated into precision agricultural systems to assist in the planning of routes for loaders and bale wagons. The approach is experimentally tested on a simulated study area like those found in real situations.

Unsupervised Cryo-EM Data Clustering through Adaptively Constrained K-Means Algorithm.

In single-particle cryo-electron microscopy (cryo-EM), K-means clustering algorithm is widely used in unsupervised 2D classification of projection images of biological macromolecules. 3D ab initio reconstruction requires accurate unsupervised classification in order to separate molecular projections of distinct orientations. Due to background noise in single-particle images and uncertainty of molecular orientations, traditional K-means clustering algorithm may classify images into wrong classes and produce classes with a large variation in membership. Overcoming these limitations requires further development on clustering algorithms for cryo-EM data analysis. We propose a novel unsupervised data clustering method building upon the traditional K-means algorithm. By introducing an adaptive constraint term in the objective function, our algorithm not only avoids a large variation in class sizes but also produces more accurate data clustering. Applications of this approach to both simulated and experimental cryo-EM data demonstrate that our algorithm is a significantly improved alterative to the traditional K-means algorithm in single-particle cryo-EM analysis.

Real-time identification of probe vehicle trajectories in the mixed traffic corridor

This paper proposes three enhanced semi-supervised clustering algorithms, namely the Constrained-K-Means (CKM), the Seeded-K-Means (SKM), and the Semi-Supervised Fuzzy c-Means (SFCM), to identify probe vehicle trajectories in the mixed traffic corridor. The proposed algorithms are able to take advantage of the strengthens of topological relation judgment and the semi-supervised learning technique by optimizing the selection of pre-labeling samples and initial clustering centers of the original semi-supervised learning technique based on horizontal Global Positioning System data. The proposed algorithms were validated and evaluated based on the probe vehicle data collected at two mixed corridors on Shanghai’s urban expressways. Results indicate that the enhanced SFCM algorithm could achieve the best performance in terms of clustering purity and Normalized Mutual Information, followed by the CKM algorithm and the SKM algorithm. It may reach a nearly 100% clustering purity for the uncongested conditions and a clustering purity greater than 80% for the congested conditions. Meanwhile, it could improve clustering purity averagely by 21% and 14% for the congested conditions and 6.5% and 6% for the uncongested conditions, as compared with the traditional K-Means algorithm and the basic SFCM. The proposed algorithms can be applied for both on-line and off-line purposes, without the need of historical data. Clustering accuracies under different traffic conditions and possible improvements with the use of historical data are also discussed.

Computationally Efficient Simulation of Multicomponent Fuel Combustion Using a Sparse Analytical Jacobian Chemistry Solver and High-Dimensional Clustering

The need for more efficient and environmentally sustainable internal combustion engines is driving research towards the need to consider more realistic models for both fuel physics and chemistry. As far as compression ignition engines are concerned, phenomenological or lumped fuel models are unreliable to capture spray and combustion strategies outside of their validation domains—typically, high-pressure injection and high-temperature combustion. Furthermore, the development of variable-reactivity combustion strategies also creates the need to model comprehensively different hydrocarbon families even in single fuel surrogates. From the computational point of view, challenges to achieving practical simulation times arise from the dimensions of the reaction mechanism, which can be of hundreds species even if hydrocarbon families are lumped into representative compounds and, thus, modeled with nonelementary, skeletal reaction pathways. In this case, it is also impossible to pursue further mechanism reductions to lower dimensions. central processing unit (CPU) times for integrating chemical kinetics in internal combustion engine simulations ultimately scale with the number of cells in the grid and with the cube number of species in the reaction mechanism. In the present work, two approaches to reduce the demands of engine simulations with detailed chemistry are presented. The first one addresses the demands due to the solution of the chemistry ordinary differential equation (ODE) system, and features the adoption of SpeedCHEM, a newly developed chemistry package that solves chemical kinetics using sparse analytical Jacobians. The second one aims to reduce the number of chemistry calculations by binning the computational fluid dynamics (CFD) cells of the engine grid into a subset of clusters, where chemistry is solved and then mapped back to the original domain. In particular, a high-dimensional representation of the chemical state space is adopted for keeping track of the different fuel components, and a newly developed bounding-box- constrained k-means algorithm is used to subdivide the cells into reactively homogeneous clusters. The approaches have been tested on a number of simulations featuring multicomponent diesel fuel surrogates and different engine grids. The results show that significant CPU time reductions, of about 1 order of magnitude, can be achieved without loss of accuracy in both engine performance and emissions predictions, prompting for their applicability to more refined or full-sized engine grids.