Trade-Off between Efficiency and Effectiveness: A Late Fusion Multi-View Clustering Algorithm

Late fusion multi-view clustering (LFMVC) algorithms aim to integrate the base partition of each single view into a consensus partition. Base partitions can be obtained by performing kernel k-means clustering on all views. This type of method is not only computationally efficient, but also more accurate than multiple kernel k-means, and is thus widely used in the multi-view clustering context. LFMVC improves computational efficiency to the extent that the computational complexity of each iteration is reduced from O(n³) to O(n) (where n is the number of samples). However, LFMVC also limits the search space of the optimal solution, meaning that the clustering results obtained are not ideal. Accordingly, in order to obtain more information from each base partition and thus improve the clustering performance, we propose a new late fusion multi-view clustering algorithm with a computational complexity of O(n²). Experiments on several commonly used datasets demonstrate that the proposed algorithm can reach quickly convergence. Moreover, compared with other late fusion algorithms with computational complexity of O(n), the actual time consumption of the proposed algorithm does not significantly increase. At the same time, comparisons with several other state-of-the-art algorithms reveal that the proposed algorithm also obtains the best clustering performance.     

Fuzzy C-Means Algorithm Automatically Determining Optimal Number of Clusters

In clustering analysis, the key to deciding clustering quality is to determine the optimal number of clusters. At present, most clustering algorithms need to give the number of clusters in advance for clustering analysis of the samples. How to gain the correct optimal number of clusters has been an important topic of clustering validation study. By studying and analyzing the FCM algorithm in this study, an accurate and efficient algorithm used to confirm the optimal number of clusters is proposed for the defects of traditional FCM algorithm. For time and clustering accuracy problems of FCM algorithm and relevant algorithms automatically determining the optimal number of clusters, kernel function, AP algorithm and new evaluation indexes were applied to improve the confirmation of complexity and search the scope of traditional fuzzy C-means algorithm, and evaluation of clustering results. Besides, three groups of contrast experiments were designed with different datasets for verification. The results showed that the improved algorithm improves time efficiency and accuracy to certain degree.     

量子蚁群模糊聚类算法在图像分割中的应用

针对模糊C-均值算法对初始值的依赖,容易陷入局部最优值的缺点,本文提出将量子蚁群算法与FCM聚类算法结合,首先利用量子蚁群算法的全局性和鲁棒性以及快速收敛的优点确定图像的初始聚类中心和聚类个数,再将所得结果作为FCM聚类算法的初始参数,然后用FCM聚类算法对医学图像进行分割。实验结果表明,该方法有效解决了FCM算法对初始参数的依赖,克服了FCM算法及蚁群算法容易陷入局部极值的的缺点,而且在分割速度和精度上得到了较大提高。     

A Direct Data-Cluster Analysis Method Based on Neutrosophic Set Implication

Raw data are classified using clustering techniques in a reasonable manner to create disjoint clusters. A lot of clustering algorithms based on specific parameters have been proposed to access a high volume of datasets. This paper focuses on cluster analysis based on neutrosophic set implication, i.e., a k-means algorithm with a threshold-based clustering technique. This algorithm addresses the shortcomings of the k-means clustering algorithm by overcoming the limitations of the threshold-based clustering algorithm. To evaluate the validity of the proposed method, several validity measures and validity indices are applied to the Iris dataset (from the University of California, Irvine, Machine Learning Repository) along with k-means and threshold-based clustering algorithms. The proposed method results in more segregated datasets with compacted clusters, thus achieving higher validity indices. The method also eliminates the limitations of threshold-based clustering algorithm and validates measures and respective indices along with k-means and thresholdbased clustering algorithms.     

基于模糊聚类的光电经纬仪多子弹弹道测量

本文对密集多子弹在空间坐标系中的左右相机弹点分布识别提出了一种新的方法。在各帧拍摄所得多子弹成像点中,提取子弹的形心坐标,根据子弹运动特点建立状态方程和测量方程,利用Kalman算法进行滤波。利用模糊c-均值聚类算法确定测量与弹丸轨迹的关联程度,从而实现弹道识别。仿真实验结果说明,该方法能有效地进行多子弹的弹道识别跟踪,并且算法简单、计算量小、易于工程实现。     

Colour quantisation using the adaptive distributing units algorithm

Colour quantisation (CQ) is an important operation with many applications in graphics and image processing. Most CQ methods are essentially based on data clustering algorithms one of which is the popular k-means algorithm. Unfortunately, like many batch clustering algorithms, k-means is highly sensitive to the selection of the initial cluster centres. In this paper, we adapt Uchiyama and Arbib’s competitive learning algorithm to the CQ problem. In contrast to the batch k-means algorithm, this online clustering algorithm does not require cluster centre initialisation. Experiments on a diverse set of publicly available images demonstrate that the presented method outperforms some of the most popular quantisers in the literature.     

基于Dirichlet过程的无线视频码率变化识别算法

研究了通过对终端视频帧质量的聚类分析来识别无线视频传输中码率变化的方法,以便为无线视频传输过程中视频码率自适应调整提供参考依据.针对经典模糊C均值(FCM)算法和K均值(K-means)算法需要设定聚类数目的问题,提出一种基于荻利克雷过程(DP)的FCM算法——DP-FCM算法.该算法将Dirichlet过程和FCM算法相结合,由视频帧信息权重峰值信噪比(IWPSNR)值使用DP过程混合模型模拟估计出聚类数目,然后进行FCM模糊聚类,通过设定合理的阈值,合并聚类结果相似项,完成视频帧的聚类,从而实现视频传输码率变化的识别.以LIVE视频库为试验数据源,对该算法进行了性能测试.试验结果表明,DP-FCM算法能够在无需设定聚类数目的前提下实现视频传输码率变化的分类识别.     

MRI brain image classification—a hybrid approach

The present article proposes a novel computer‐aided diagnosis (CAD) technique for the classification of the magnetic resonance brain images. The current method adopt color converted hybrid clustering segmentation algorithm with hybrid feature selection approach based on IGSFFS (Information gain and Sequential Forward Floating Search) and Multi‐Class Support Vector Machine (MC‐SVM) classifier technique to segregate the magnetic resonance brain images into three categories namely normal, benign and malignant. The proposed hybrid evolutionary segmentation algorithm which is the combination of WFF(weighted firefly) and K‐means algorithm called WFF‐K‐means and modified cuckoo search (MCS) and K‐means algorithm called MCS‐K‐means, which can find better cluster partition in brain tumor datasets and also overcome local optima problems in K‐means clustering algorithm. The experimental results show that the performance of the proposed algorithm is better than other algorithms such as PSO‐K‐means, color converted K‐means, FCM and other traditional approaches. The multiple feature set comprises color, texture and shape features derived from the segmented image. These features are then fed into a MC‐SVM classifier with hybrid feature selection algorithm, trained with data labeled by experts, enabling the detection of brain images at high accuracy levels. The performance of the method is evaluated using classification accuracy, sensitivity, specificity, and receiver operating characteristic (ROC) curves. The proposed method provides highest classification accuracy of greater than 98% with high sensitivity and specificity rates of greater than 95% for the proposed diagnostic model and this shows the promise of the approach. © 2015 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 25, 226–244, 2015     

基于粒子群优化聚类的汽轮机组振动故障诊断

针对模糊C-均值聚类算法(FCM)容易陷入局部极值和对初始值敏感的不足,提出了一种新的模糊聚类算法(PFCM),新算法利用粒子群优化算法(PSO)全局寻优、快速收敛的特点,代替了FCM算法的基于梯度下降的迭代过程,使算法具有很强的全局搜索能力,很大程度上避免了FCM算法易陷入局部极值的缺陷,同时也降低了FCM算法对初始值的敏感度。将该算法应用于汽轮机组振动故障诊断中,与电厂运行实际故障状态对照,仿真结果表明该算法提高了故障诊断的正确率。为汽轮机振动故障诊断方法的研究提供了一种新的思路。     

基于模糊c均值聚类算法的控制图模式识别

控制图模式识别能够区分制造过程中的一般因素与异常因素,提高制造过程中的产品质量,减少成本,提高效益。利用蒙特卡洛方法产生样本;采用一维离散小波变换处理原始数据;利用模糊c均值聚类算法进行控制图模式识别。识别准确率99.43%,其标准差为0.002 8。这表明基于该方法的控制图模式识别准确率高,稳定性好,较现有的控制图模式识别方法具有简易、高效等特点。     

Investigation of self-organizing oscillator networks for use in clustering microarray data

The self-organizing oscillator network (SOON) is a comparatively new clustering algorithm that does not require the knowledge of the number of clusters. The SOON is distance based, and its clustering behavior is different to density-based algorithms in a number of ways. This paper examines the effect of adjusting the control parameters of the SOON with four different datasets; the first is a (communications) modulation dataset representing one modulation scheme under a variety of noise conditions. This allows the assessment of the behavior of the algorithm with data varying between highly separable and nonseparable cases. The main thrust of this paper is to evaluate its efficacy in biological datasets. The second is taken from microarray experiments on the cell cycle of yeast, while the third and the fourth represent two microarray cancer datasets, i.e., the lymphoma and the liver cancer datasets. The paper demonstrates that the SOON is a viable tool to analyze these problems, and can add many useful insights to the biological data that may not always be available using other clustering methods.     

Adaptive Fuzzy C-Means clustering in process monitoring

Quite often, quality control models fail because, e.g., the mean values are changing continuously. These kinds of changes, e.g., process drifts due to seasonal fluctuations, are common in an activated sludge waste-water treatment plant in Finland. Different Fuzzy C-Means (FCM) clustering algorithms were tested in order to cope with these kinds of seasonal effects. Firstly, a Principal Component Analysis (PCA) model was constructed in order to visualize the data set and reduce the dimensionality of the problem. Then, score values of the PCA were used in the FCM. The cluster centers represented the different process conditions (winter and summer seasons). Different algorithms were used to update the cluster centers or to give them some flexibility. The testing of different FCM algorithms was carried out by using a separate test set. The adaptive and the flexible FCM algorithms were compared to the basic non-adaptive FCM. For both cases, modifications are proposed and a simple strategy for updating the cluster centers is given.     

BHGSO: Binary Hunger Games Search Optimization Algorithm for Feature Selection Problem

In machine learning and data mining, feature selection (FS) is a traditional and complicated optimization problem. Since the run time increases exponentially, FS is treated as an NP-hard problem. The researcher’s effort to build a new FS solution was inspired by the ongoing need for an efficient FS framework and the success rates of swarming outcomes in different optimization scenarios. This paper presents two binary variants of a Hunger Games Search Optimization (HGSO) algorithm based on V- and S-shaped transfer functions within a wrapper FS model for choosing the best features from a large dataset. The proposed technique transforms the continuous HGSO into a binary variant using V- and S-shaped transfer functions (BHGSO-V and BHGSO-S). To validate the accuracy, 16 famous UCI datasets are considered and compared with different state-of-the-art metaheuristic binary algorithms. The findings demonstrate that BHGSO-V achieves better performance in terms of the selected number of features, classification accuracy, run time, and fitness values than other state-of-the-art algorithms. The results demonstrate that the BHGSO-V algorithm can reduce dimensionality and choose the most helpful features for classification problems. The proposed BHGSO-V achieves 95% average classification accuracy for most of the datasets, and run time is less than 5 sec. for low and medium dimensional datasets and less than 10 sec for high dimensional datasets.     

A new hybrid teaching–learning particle swarm optimization algorithm for synthesis of linkages to generate path

This paper proposes a novel hybrid teaching–learning particle swarm optimization (HTLPSO) algorithm, which merges two established nature-inspired algorithms, namely, optimization based on teaching–learning (TLBO) and particle swarm optimization (PSO). The HTLPSO merges the best half of population obtained after the teacher phase in TLBO with the best half of the population obtained after PSO. The population so obtained is used subsequently in learner phase of TLBO. To validate the proposed algorithm, five constrained benchmark functions are considered to prove its robustness and efficiency. The proposed algorithm is applied to synthesize four-bar linkage for prescribed path. It is found that the HTLPSO performs better than other single nature-inspired algorithms for path synthesis problem in mechanism theory. Hence, HTLPSO may prove to be an important tool for mechanism design to follow the prescribed path.     

Collaborative Filtering for Massive Datasets Based on Bayesian Networks

This paper proposes a collaborative filtering method for massive datasets that is based on Bayesian networks. We first compare the prediction accuracy of four scoring-based learning Bayesian networks algorithms (AIC, MDL, UPSM, and BDeu) and two conditional-independence-based (Cl-based) learning Bayesian networks algorithms (MWST, and Polytree-MWST) using actual massive datasets. The results show that (1) for large networks, the scoring-based algorithms have lower prediction accuracy than the CI-based algorithms and (2) when the scoring-based algorithms use a greedy search to learn a large network, algorithms which make a lot of arcs tend to have less prediction accuracy than those that make fewer arcs. Next, we propose a learning algorithm based on MWST for collaborative filtering of massive datasets. The proposed algorithm employs a traditional data mining technique, the “a priori” algorithm, to quickly calculate the amount of mutual information, which is needed in MWST, from massive datasets. We compare the original MWST algorithm and the proposed algorithm on actual data, and the comparison shows the effectiveness of the proposed algorithm.

     

Modeling Bacterial Species: Using Sequence Similarity with Clustering Techniques

Existing studies have challenged the current definition of named bacterial species, especially in the case of highly recombinogenic bacteria. This has led to considering the use of computational procedures to examine potential bacterial clusters that are not identified by species naming. This paper describes the use of sequence data obtained from MLST databases as input for a k-means algorithm extended to deal with housekeeping gene sequences as a metric of similarity for the clustering process. An implementation of the k-means algorithm has been developed based on an existing source code implementation, and it has been evaluated against MLST data. Results point out to potential bacterial clusters that are close to more than one different named species and thus may become candidates for alternative classifications accounting for genotypic information. The use of hierarchical clustering with sequence comparison as similarity metric has the potential to find clusters different from named species by using a more informed cluster formation strategy than a conventional nominal variant of the algorithm.     

Performance analysis of classifiers for colon cancer detection from dimensionality reduced microarray gene data

Cancer disease is accountable for many deaths that are over 9.6 million in 2018 and roughly one out of six deaths occur because of cancer worldwide. The colon cancer is the second prominent source of death of around 1.8 million cases. This research is inclined to detect the colon cancer from microarray dataset. It will aids the experts to distinguish the cancer cells from normal cells for appropriate determination and treatment of cancer at earlier stages that leads to increase the survival rate of the patients. The high dimensionality in microarray dataset with less samples and more attributes creates lag in the detection capability of the classifier. Hence there is a need for dimensionality reduction techniques to preserve the significant genes that are prominent in the disease classification. In this article, at first ANOVA method used to select the best genes and then principal component analysis (PCA) and fuzzy C-means clustering (FCM) techniques are further employed to choose relevant genes. The PCA and FCM features are classified using model, discriminant, regression, hybrid, and heuristic-based classifiers. The attained results show that the heuristic classifier with PCA features is encapsulated an average classification accuracy of 97.92% for classifying both the colon cancer and normal samples. Also, for FCM features, the Heuristic classifier is maintained at an average classification accuracy of 99.48% and 97.92% for classifying the colon cancer and normal samples, respectively. The Heuristic classifier outperforms with high accuracy than all other classifiers in the classification of colon cancer.     

基于改进YOLOv4算法的轮毂表面缺陷检测

汽车轮毂加工过程中产生的表面缺陷严重影响整车的美观性及服役性能,针对人工检测效率低、漏检率高的问题,提出一种基于改进YOLOv4算法的轮毂表面缺陷检测方法。构建了轮毂缺陷数据集,其包含6种表面缺陷,由2346张4928×3264pixel的图像组成;采用K-means方法进行先验框聚类,并针对YOLOv4算法在纤维、粘铝等小尺度缺陷上检测精度不足问题,在原网络Neck部分引入细化U型网络模块(TUM)和注意力机制,用于增强有效特征并抑制无效特征,强化多尺度特征提取与融合,改善特征处理过程中可能存在的小目标信息丢失问题;基于该数据集,训练并测试不同算法的缺陷检测性能并验证改进模块的有效性。结果表明,该方法大幅提升了粘铝等小尺寸缺陷的检测能力,缺陷检测平均精度达到85.8%,与多种算法相比较检测精度最高。     

Fault Pattern Recognition based on Kernel Method and Fuzzy C-means

A method about fault identification is proposed to solve the relationship among fault features of large rotating machinery,which is extremely complicated and nonlinear. This paper studies the rotor test-rig and the clustering of data sets and fault pattern recognitions. The present method firstly maps the data from their original space to a high dimensional Kernel space which makes the highly nonlinear data in low-dimensional space become linearly separable in Kernel space. It highlights the differences among the features of the data set. Then fuzzy C-means( FCM) is conducted in the Kernel space. Each data is assigned to the nearest class by computing the distance to the clustering center. Finally,test set is used to judge the results. The convergence rate and clustering accuracy are better than traditional FCM. The study shows that the method is effective for the accuracy of pattern recognition on rotating machinery.     

A new unified theory underlying time dependent linear first‐order systems: a prelude to algorithms by design

A new unified theory underlying the theoretical design of linear computational algorithms in the context of time dependent first‐order systems is presented. Providing for the first time new perspectives and fresh ideas, and unlike various formulations existing in the literature, the present unified theory involves the following considerations: (i) it leads to new avenues for designing new computational algorithms to foster the notion of algorithms by design and recovering existing algorithms in the literature, (ii) describes a theory for the evolution of time operators via a unified mathematical framework, and (iii) places into context and explains/contrasts future new developments including existing designs and the various relationships among the different classes of algorithms in the literature such as linear multi‐step methods, sub‐stepping methods, Runge–Kutta type methods, higher‐order time accurate methods, etc. Subsequently, it provides design criteria and guidelines for contrasting and evaluating time dependent computational algorithms. The linear computational algorithms in the context of first‐order systems are classified as distinctly pertaining to Type 1, Type 2, and Type 3 classifications of time discretized operators. Such a distinct classification, provides for the first time, new avenues for designing new computational algorithms not existing in the literature and recovering existing algorithms of arbitrary order of time accuracy including an overall assessment of their stability and other algorithmic attributes. Consequently, it enables the evaluation and provides the relationships of computational algorithms for time dependent problems via a standardized measure based on computational effort and memory usage in terms of the resulting number of equation systems and the corresponding number of system solves. A generalized stability and accuracy limitation barrier theorem underlies the generic designs of computational algorithms with arbitrary order of accuracy and establishes guidelines which cannot be circumvented. In summary, unlike the traditional approaches and classical school of thought customarily employed in the theoretical development of computational algorithms, the unified theory underlying time dependent first‐order systems serves as a viable avenue to foster the notion of algorithms by design. Copyright © 2004 John Wiley & Sons, Ltd.