Enhance Clustering Performance Research Articles

Incomplete multi-view subspace clustering based on missing-sample recovering and structural information learning

The incomplete multi-view clustering (IMVC) focuses on exploring hidden information from missing samples and complementary information from multiple views to enhance clustering performance. Most of existing methods for IMVC only attempt to obtain a shared consistent representation or graph from the non-missing samples based on matrix factorization. However, these methods rely too much on view alignment information, neglecting the specific-view structural information of different views and the hidden information of missing samples. To address the above problems, a novel method, called incomplete multi-view subspace clustering based on missing-sample recovering and structural information learning (MSR_SIL), is proposed in this paper. Overall, MSR_SIL incorporates the missing-sample recovering in feature space and self-expressiveness coefficient learning in representation space into a unified framework. Specifically, a reconstruction term is introduced to recover missing samples with the help of non-missing ones so that incomplete multi-view is completed. Next, the self-expressiveness coefficient of the recovered complete multi-view is learned. Then, each self-expressiveness coefficient is decomposed into a consistent part to capture the similarity information of view-paired and a specific part to hold the unique information of each view. Especially, to retain the global manifold structure of the multi-view data, Schatten-p norm is applied. Finally, the clustering results can be obtained on the basis of the consistent and specific representation. Experimental results on eleven well-known benchmark multi-view datasets demonstrate that the proposed MSR_SIL yields significant improvements on clustering performance.

Query-Augmented Active Metric Learning

In this article, we propose an active metric learning method for clustering with pairwise constraints. The proposed method actively queries the label of informative instance pairs, while estimating underlying metrics by incorporating unlabeled instance pairs, which leads to a more accurate and efficient clustering process. In particular, we augment the queried constraints by generating more pairwise labels to provide additional information in learning a metric to enhance clustering performance. Furthermore, we increase the robustness of metric learning by updating the learned metric sequentially and penalizing the irrelevant features adaptively. In addition, we propose a novel active query strategy that evaluates the information gain of instance pairs more accurately by incorporating the neighborhood structure, which improves clustering efficiency without extra labeling cost. In theory, we provide a tighter error bound of the proposed metric learning method using augmented queries compared with methods using existing constraints only. Furthermore, we also investigate the improvement using the active query strategy instead of random selection. Numerical studies on simulation settings and real datasets indicate that the proposed method is especially advantageous when the signal-to-noise ratio between significant features and irrelevant features is low.

Multiview Common Subspace Clustering via Coupled Low Rank Representation

Multi-view subspace clustering (MVSC) finds a shared structure in latent low-dimensional subspaces of multi-view data to enhance clustering performance. Nonetheless, we observe that most existing MVSC methods neglect the diversity in multi-view data by considering only the common knowledge to find a shared structure either directly or by merging different similarity matrices learned for each view. In the presence of noise, this predefined shared structure becomes a biased representation of the different views. Thus, in this article, we propose a MVSC method based on coupled low-rank representation to address the above limitation. Our method first obtains a low-rank representation for each view, constrained to be a linear combination of the view-specific representation and the shared representation by simultaneously encouraging the sparsity of view-specific one. Then, it uses the k -block diagonal regularizer to learn a manifold recovery matrix for each view through respective low-rank matrices to recover more manifold structures from them. In this way, the proposed method can find an ideal similarity matrix by approximating clustering projection matrices obtained from the recovery structures. Hence, this similarity matrix denotes our clustering structure with exactly k connected components by applying a rank constraint on the similarity matrix’s relaxed Laplacian matrix to avoid spectral post-processing of the low-dimensional embedding matrix. The core of our idea is such that we introduce dynamic approximation into the low-rank representation to allow the clustering structure and the shared representation to guide each other to learn cleaner low-rank matrices that would lead to a better clustering structure. Therefore, our approach is notably different from existing methods in which the local manifold structure of data is captured in advance. Extensive experiments on six benchmark datasets show that our method outperforms 10 similar state-of-the-art compared methods in six evaluation metrics.

Fuzzy Density Peaks Clustering

As an exemplar-based clustering method, the well-known density peaks clustering (DPC) heavily depends on the computation of kernel-based density peaks, which incurs two issues: first, whether kernel-based density can facilitate a large variety of data well, including cases where ambiguity and uncertainty of the assignment of the data points to their clusters may exist, and second, whether the concept of density peaks can be interpreted and manipulated from the perspective of soft partitions (e.g., fuzzy partitions) to achieve enhanced clustering performance. In this article, in order to provide flexible adaptability for tackling ambiguity and uncertainty in clustering, a new concept of fuzzy peaks is proposed to express the density of a data point as the fuzzy-operator-based coupling of the fuzzy distances between a data point and its neighbors. As a fuzzy variant of DPC, a novel fuzzy density peaks clustering (FDPC) method FDPC based on fuzzy operators (especially S-norm operators) is accordingly devised along with the same algorithmic framework of DPC. With an appropriate choice of a fuzzy operator with its associated tunable parameter for a clustering task, FDPC can indeed inherit the advantage of fuzzy partitions and simultaneously provide flexibility in enhancing clustering performance. The experimental results on both synthetic and real data sets demonstrate that the proposed method outperforms or at least remains comparable to the comparative methods in clustering performance by choosing appropriate parameters in most cases.

An Active Learning Algorithm Based on Shannon Entropy for Constraint-Based Clustering

Pairwise constraints could enhance clustering performance in constraint-based clustering problems, especially when these pairwise constraints are informative. In this paper, a novel active learning pairwise constraint formulation algorithm would be constructed with aim to formulate informative pairwise constraints efficiently and economically. This algorithm consists of three phases: Selecting, Exploring and Consolidating. In Selecting phase, some type of unsupervised clustering algorithm is used to obtain an informative data set in terms of Shannon entropy. In Exploring phase, some type of farthest-first strategy is used to construct a series of query with aim to construct clustering skeleton set structure and informative pairwise constraints are also collected meanwhile based on the informative data set. If the number of skeleton sets equals the number of clusters, the new algorithm gets into third phase Consolidating; otherwise, it would finish. In Consolidating phase, non-skeleton points included in the informative data set are used to construct a series of query with skeleton set representative points constructed in Exploring phase. And some type of priority principle is proposed to help collect more must-link pairwise constraints. Treat the well-known MPCK-means (metric pairwise constrained K-means) as the underlying constraint-based semi-supervised clustering algorithm and data experiment comparison between this new algorithm and its counterparts would be done. Experiment outcome shows that significant improvement of this new algorithm.

Self-paced and auto-weighted multi-view clustering

Multi-view clustering (MVC) methods are effective approaches to enhance clustering performance by exploiting complementary information from multiple views. One main disadvantage of most existing MVC methods is that the corresponding optimization problems are non-convex and thus local optimal solutions are usually obtained. To alleviate this issue, we propose a novel multi-view clustering method equipped with self-paced learning, which first learns the MVC model with easy examples and then progressively considers complex ones from each view. In addition, a soft weighting scheme of self-paced learning is designed to further reduce the negative impact from outliers and noises. Furthermore, to consider the importance of different views, we develop an auto-weighted technique to automatically assign weights to views. The proposed model only needs the number of clusters as input and can be easily solved by an alternating optimization paradigm. Experimental results on various benchmark data sets demonstrate the effectiveness of the proposed model.

Adaptive Sample-level Graph Combination for Partial Multiview Clustering.

Multiview clustering explores complementary information among distinct views to enhance clustering performance under the assumption that all samples have complete information in all available views. However, this assumption does not hold in many real applications, where the information of some samples in one or more views may be missing, leading to partial multiview clustering problems. In this case, significant performance degeneration is usually observed. A collection of partial multiview clustering algorithms has been proposed to address this issue and most treat all different views equally during clustering. In fact, because different views provide features collected from different angles/feature spaces, they might play different roles in the clustering process. With the diversity of different views considered, in this study, a novel adaptive method is proposed for partial multiview clustering by automatically adjusting the contributions of different views. The samples are divided into complete and incomplete sets, while a joint learning mechanism is established to facilitate the connection between them and thereby improve clustering performance. More specifically, the method is characterized by a joint optimization model comprising two terms. The first term mines the underlying cluster structure from both complete and incomplete samples by adaptively updating their importance in all available views. The second term is designed to group all data with the aid of the cluster structure modeled in the first term. These two terms seamlessly integrate the complementary information among multiple views and enhance the performance of partial multiview clustering. Experimental results on real-world datasets illustrate the effectiveness and efficiency of our proposed method.

Learning Transferable Subspace for Human Motion Segmentation

Temporal data clustering is a challenging task. Existing methods usually explore data self-representation strategy, which may hinder the clustering performance in insufficient or corrupted data scenarios. In real-world applications, we are easily accessible to a large amount of related labeled data. To this end, we propose a novel transferable subspace clustering approach by exploring useful information from relevant source data to enhance clustering performance in target temporal data. We manage to transform the original data into a shared low-dimensional and distinctive feature space by jointly seeking an effective domain-invariant projection. In this way, the well-labeled source knowledge can help obtain a more discriminative target representation. Moreover, a graph regularizer is designed to incorporate temporal information to preserve more sequence knowledge into the learned representation. Extensive experiments based on three human motion datasets illustrate that our approach is able to outperform state-of-the-art temporal data clustering methods.

Enhancing Clustering Performance through Filter based Data Reduction

Clustering is an unsupervised learning method used to identify inherent grouping of set of unlabeled data. Such set of groups are termed as Clusters. Grouping of datasets in to clusters involves minimizes the interclass similarity and maximizes the intraclass similarity. Using clustering method the Data Reduction becomes a simple process of identifying a relevant subset in a very big datasets. A comparative study is carried out to identify the performance of filters on different datasets. The various clustering algorithms are tested on those datasets, the results are recorded and analyzed. The Filter is a preprocessing tool which performs certain transformations on the input data, and make the data suitable for specific application. In this paper the various clustering methods along with various datasets are tested to identify the usage of various filters performance. The recorded results proves the K-Means is the best clustering algorithm when compared with Expectation Maximization algorithm.

Time series k-means: A new k-means type smooth subspace clustering for time series data

Existing clustering algorithms are weak in extracting smooth subspaces for clustering time series data. In this paper, we propose a new k-means type smooth subspace clustering algorithm named Time Series k-means (TSkmeans) for clustering time series data. The proposed TSkmeans algorithm can effectively exploit inherent subspace information of a time series data set to enhance clustering performance. More specifically, the smooth subspaces are represented by weighted time stamps which indicate the relative discriminative power of these time stamps for clustering objects. The main contributions of our work include the design of a new objective function to guide the clustering of time series data and the development of novel updating rules for iterative cluster searching with respect to smooth subspaces. Based on a synthetic data set and five real-life data sets, our experimental results confirm that the proposed TSkmeans algorithm outperforms other state-of-the-art time series clustering algorithms in terms of common performance metrics such as Accuracy, Fscore, RandIndex, and Normal Mutual Information.

CGC: A Flexible and Robust Approach to Integrating Co-Regularized Multi-Domain Graph for Clustering.

Multi-view graph clustering aims to enhance clustering performance by integrating heterogeneous information collected in different domains. Each domain provides a different view of the data instances. Leveraging cross-domain information has been demonstrated an effective way to achieve better clustering results. Despite the previous success, existing multi-view graph clustering methods usually assume that different views are available for the same set of instances. Thus instances in different domains can be treated as having strict one-to-one relationship. In many real-life applications, however, data instances in one domain may correspond to multiple instances in another domain. Moreover, relationships between instances in different domains may be associated with weights based on prior (partial) knowledge. In this paper, we propose a flexible and robust framework, CGC (Co-regularized Graph Clustering), based on non-negative matrix factorization (NMF), to tackle these challenges. CGC has several advantages over the existing methods. First, it supports many-to-many cross-domain instance relationship. Second, it incorporates weight on cross-domain relationship. Third, it allows partial cross-domain mapping so that graphs in different domains may have different sizes. Finally, it provides users with the extent to which the cross-domain instance relationship violates the in-domain clustering structure, and thus enables users to re-evaluate the consistency of the relationship. We develop an efficient optimization method that guarantees to find the global optimal solution with a given confidence requirement. The proposed method can automatically identify noisy domains and assign smaller weights to them. This helps to obtain optimal graph partition for the focused domain. Extensive experimental results on UCI benchmark data sets, newsgroup data sets and biological interaction networks demonstrate the effectiveness of our approach.

A fuzzy logic-based clustering algorithm for network optimisation

Rate of occurrence of high-dimensional data is much higher and sad to relate, classical clustering techniques do not hold good for such high-dimensional networks of arbitrary shapes in the underwater wireless sensor network. This is mainly due to the fact that clustering techniques are highly parameterised. Data aggregation using traditional clustering techniques is a problem for high-dimensional network because of its arbitrary shapes. Moreover, existing clustering algorithms cannot be used for solving the problem and enhancing clustering performance. Numerous top-notch researchers have employed fuzzy logic-based clustering schemes in the past. We analyse these techniques to find scientific tendency. In this paper, a fuzzy-based algorithm for clustering ensemble is proposed where sensor nodes are characterised based upon major and minor criteria. This algorithm is developed to group the sensor nodes in multiple clusters. The clustering precision is designed to assess the efficacy of the algorithm. This algorithm also competes in its ability to distinguish the difference between all the sensor nodes present in the network. Other clustering algorithms in existence for data aggregation are also reviewed and compared with the proposed technique.

Subspace Clustering of Text Documents Using Collection and Document Frequencies of Terms

The most widely used document representation model is the Vector Space Model. Higher dimensions and sparseness of the representation model leads to poor clustering performance, demanding more computational effort for clustering. Hence, dimension reduction techniques are used to find a feature subspace for document representation that could enhance clustering performance. This paper proposes a novel unsupervised filter method for feature selection. Feature selection methods represent documents using a subset of the original feature set that maximizes the separation among classes of documents in the collection. Filter methods analyze the intrinsic properties of the documents and they select highly-ranked features according to some criterion, quite different to clustering task. Unsupervised feature selection methods do not use class labels to guide the selection of features. The proposed method assigns a score to a term using its collection and document frequencies. Number of times and number of documents in which a term appears in a document collection are called respectively collection frequency and document frequency of the term. Empirical evaluations proved that the proposed method is not only effective in selecting features giving best clustering performance, but also less computationally complex, when compared to other unsupervised feature selection methods.

Improved suffix tree clustering for Uyghur text

In order to solve the problems of non-standard,repetition and redundancy of information in the process of selecting the base class phrases,an improved Suffix Tree Clustering(STC) method was proposed.Firstly,phrase mutual information algorithm was put forward to choose the base class phrases abiding by Uyghur grammar.Secondly,in order to reduce the repeated base class phrase,the phrase reduction algorithm based on Uyghur grammar was proposed.Thirdly,on the basis of the first two steps,the phrase redundancy algorithm based on Uyghur grammar was constructed to remove redundant phrase.The experimental results show that this method improves the recall and the precision compared with STC.This indicates that the improved algorithm can enhance clustering performance effectively.

Aggregate two-way co-clustering of ads and user data for online advertisements

Clustering plays an important role in data mining, as it is used by many applications as a preprocessing step for data analysis. Traditional clustering focuses on grouping similar objects, while two-way co-clustering can group dyadic data (objects as well as their attributes) simultaneously. In this research, we apply two-way co-clustering to the analysis of online advertising where both ads and users need to be clustered. However, in addition to the ad-user link matrix that denotes the ads which a user has linked, we also have two additional matrices, which represent extra information about users and ads. In this paper, we proposed a 3-staged clustering method that makes use of the three data matrices to enhance clustering performance. In addition, an Iterative Cross Co-Clustering (ICCC) algorithm is also proposed for two-way co-clustering. The experiment is performed using the advertisement and user data from Morgenstern, a financial social website that focuses on the agency of advertisements. The result shows that iterative cross co-clustering provides better performance than traditional clustering and completes the task more efficiently.

Tri-Training and Data Editing Based Semi-Supervised Clustering Algorithm

In this paper, a algorithm named DE-Tri-training semi-supervised K-means is proposed, which could get a seeds set of larger scale and less noise. In detail, prior to using the seeds set to initialize cluster centroids, the training process of a semi-supervised classification approach named Tri-training is used to label unlabeled data and add them into the initial seeds set to enlarge the scale. Meanwhile, to improve the quality of the enlarged seeds set, a nearest neighbor rule based data editing technique named Depuration is introduced into Tri-training process to eliminate and correct the mislabeled noise data in the enlarged seeds. Experimental results show that the novel semi-supervised clustering algorithm could effectively improve the cluster centroids initialization and enhance clustering performance.