Single Linkage Hierarchical Clustering Research Articles

The effect of clustering algorithms on question answering

Question answering (QA) is one of the essential fields in information retrieval where specific answers are provided instead of large documents. The relations among questions and answers are determined using natural language processing techniques while clustering algorithms can be helpful in improving the effectiveness of result retrieval by reducing the amount of required comparisons for a specific question or answer. In this work, we introduce a clustering-based approach for a QA system. This approach groups related questions into clusters using different clustering algorithms, specifies the appropriate answer using similarity methods between the answers and the generated clusters, and then assigns answers to their most related questions. Different clustering algorithms, such as k-means, spherical k-means, single-linkage hierarchical clustering (SLHA), unweighted pair group method with arithmetic mean (UPGMA), expectation–maximization (EM), and clustering Arabic documents based on bond energy (CADBE), are tested. The effectiveness of a clustering algorithm is investigated with respect to certain factors, including number of clusters, text representation, similarity measure between answers and clusters, and similarity measure between answers and questions in a selected cluster. In addition, a comprehensive ranking system is introduced to evaluate the performance of clustering algorithms. Evaluation is performed using the Dataset of Arabic Why Question Answering System (DAWQAS) and the Multilingual Question Answering (MLQA) dataset. Results show that CADBE achieves the highest accuracy and the first rank, followed by SLHA and UPGMA, while spherical k-means has the lowest rank. The performance of clustering algorithms for MLQA dataset is affected by its characteristics, such as short questions, long and varied answers, and diverse subject domains. Unigram and bigram intersection measures perform well in most cases. Term frequency inverse document frequency representation outperforms word embedding in DAWQAS. Overall, the experiments provide insights into the performance of clustering algorithms in QA systems.

Reduced Bayesian Optimized Stacked Regressor (RBOSR): A highly efficient stacked approach for improved air pollution prediction

Over the past few years, stacked generalization has emerged as a promising approach for predicting the concentration of PM2.5, but its use is often associated with reduced efficiency due to computational demands. In this study, we propose a new approach called Reduced Bayesian Optimized Stacked Regressor (RBOSR) that optimizes both the performance and efficiency of the PM2.5 stacked model. RBOSR incorporates Bayesian optimization for hyperparameter tuning, ensemble-based feature selection, dimensionality reduction via single-link hierarchical clustering, and recursive base estimator eliminations. Results:The RBOSR model is significantly more efficient when compared to the original stacked model, with 5.7 times shorter training time. Additionally, the RBOSR model outperformed the unreduced stacked model with an R2 value of 0.91 and RMSE of 26.46, resulting in a 3.5% improvement. Compared to other PM2.5 stacked models proposed in recent studies, the RBOSR model demonstrates superior efficiency, with up to 47 times shorter training time. While the RBOSR approach has been developed specifically for PM2.5 prediction, it has the potential for broader applications in other regression problems. Although the approach has not yet been applied to other datasets, future work could explore its applicability to a broader range of datasets and the development of more efficient strategies for optimizing base estimator selection.

On parsimony and clustering.

This work is motivated by applications of parsimonious cladograms for the purpose of analyzing non-biological data. Parsimonious cladograms were introduced as a means to help understanding the tree of life, and are now used in fields related to biological sciences at large, e.g., to analyze viruses or to predict the structure of proteins. We revisit parsimonious cladograms through the lens of clustering and compare cladograms optimized for parsimony with dendograms obtained from single linkage hierarchical clustering. We show that despite similarities in both approaches, there exist datasets whose clustering dendogram is incompatible with parsimony optimization. Furthermore, we provide numerical examples to compare via F-scores the clustering obtained through both parsimonious cladograms and single linkage hierarchical dendograms.

SSTD: A Novel Spatio-Temporal Demographic Network for EEG-Based Emotion Recognition

Emotion recognition is the key to making machines more intelligent. This study proposes a novel sing-link end-to-end spatio-temporal demographic network (SSTD) that fuses spatial, temporal, and demographic information for electroencephalography (EEG)-based emotion recognition. In the SSTD model, an adaptive time window using single-link hierarchical clustering based on Riemannian metrics was realized for data preprocessing to solve the problem of individual differences. Then, the preprocessed EEG data acted as a gate recurrent unit (GRU) network input to calculate high-level time-domain features. At the same time, the EEG covariance matrices were fed into the symmetric positive definite matrix network (SPDNet) to calculate high-level spatial features. Given the correlation between EEG signals and individual demographic information, gender and age factors were integrated into the spatio-temporal model, resulting in more effective high-level features for EEG-based emotion recognition. Finally, extensive comparative experiments were conducted on two public datasets: DEAP and DREAMER. The average accuracy of valence and arousal on the DEAP dataset are 68.28% and 71.48%, respectively. The average accuracy of valence and arousal on the DREAMER dataset are 76.81% and 81.64%, respectively. Experimental results show that the SSTD model has an excellent recognition performance.

A Giant Arc on the Sky

ABSTRACT We present the serendipitous discovery of a ‘Giant Arc on the Sky’ at $z$ ∼ 0.8. The Giant Arc (GA) spans ∼1 Gpc (proper size, present epoch) and appears to be almost symmetrical on the sky. It was discovered via intervening Mg ii absorbers in the spectra of background quasars, using the catalogues of Zhu & Ménard. The use of Mg ii absorbers represents a new approach to the investigation of large-scale structures (LSSs) at redshifts $0.45 \la z \la 2.25$. We present the observational properties of the GA, and we assess it statistically using methods based on (i) single-linkage hierarchical clustering (∼4.5σ); (ii) the Cuzick-Edwards test (∼3.0σ); and (iii) power-spectrum analysis (∼4.8σ). Each of these methods has distinctive attributes and powers, and we advise considering the evidence from the ensemble. We discuss our approaches to mitigating any post hoc aspects of analysing significance after discovery. The overdensity of the GA is δρ/ρ ∼ 1.3 ± 0.3. The GA is the newest and one of the largest of a steadily accumulating set of very large LSSs that may (cautiously) challenge the Cosmological Principle, upon which the ‘standard model’ of cosmology is founded. Conceivably, the GA is the precursor of a structure like the Sloan Great Wall (but the GA is about twice the size), seen when the Universe was about half its present age.

Adaptive Seriational Risk Parity and Other Extensions for Heuristic Portfolio Construction Using Machine Learning and Graph Theory

In this article, the authors present a conceptual framework named <i>adaptive seriational risk parity</i> (ASRP) to extend hierarchical risk parity (HRP) as an asset allocation heuristic. The first step of HRP (quasi-diagonalization), determining the hierarchy of assets, is required for the actual allocation done in the second step (recursive bisectioning). In the original HRP scheme, this hierarchy is found using single-linkage hierarchical clustering of the correlation matrix, which is a static tree-based method. The authors compare the performance of the standard HRP with other static and adaptive tree-based methods, as well as seriation-based methods that do not rely on trees. Seriation is a broader concept allowing reordering of the rows or columns of a matrix to best express similarities between the elements. Each discussed variation leads to a different time series reflecting portfolio performance using a 20-year backtest of a multi-asset futures universe. Unsupervised learningbased on these time-series creates a taxonomy that groups the strategies in high correspondence to the construction hierarchy of the various types of ASRP. Performance analysis of the variations shows that most of the static tree-based alternatives to HRP outperform the single-linkage clustering used in HRP on a risk-adjusted basis. Adaptive tree methods show mixed results, and most generic seriation-based approaches underperform. <b>Key Findings</b> ▪ The authors introduce the <i>adaptive seriational risk parity</i> (ASRP) framework as a hierarchy of decisions to implement the quasi-diagonalization step of hierarchical risk parity (HRP) with seriation-based and tree-based variations as alternatives to single linkage. Tree-based variations are further separated in static and adaptive versions. Altogether, 57 variations are discussed and connected to the literature. ▪ Backtests of the 57 different HRP-type asset allocation variations applied to a multi-asset futures universe lead to a correlation matrix of the resulting 57 portfolio return time series. This portfolio return correlation matrix can be visualized as a dendrogram using single-linkage clustering. The correlation hierarchy reflected by the dendrogram is similar to the construction hierarchy of the quasi-diagonalization step. Most seriation-based strategies seem to underperform HRP on a risk-adjusted basis. Most static tree-based variations outperform HRP, whereas adaptive tree-based methods show mixed results. ▪ The presented variations fit into a triple artificial intelligence approach to connect synthetic data generation with explainable machine learning. This approach generates synthetic market data in the first step. The second step applies an HRP-type portfolio allocation approach as discussed in this article. The third step uses a model-agnostic explanation such as the SHAP framework to explain the resulting performance with features of the synthetic market data and with model selection in the second step.

'Adaptive Seriational Risk Parity' and other Extensions for Heuristic Portfolio Construction using Machine Learning and Graph Theory

In this article, the authors present a conceptual framework named 'Adaptive Seriational Risk Parity' (ASRP) to extend Hierarchical Risk Parity (HRP) as an asset allocation heuristic. The first step of HRP (quasi-diagonalization) determining the hierarchy of assets is required for the actual allocation in the second step of HRP (recursive bisectioning). In the original HRP scheme, this hierarchy is found using the single-linkage hierarchical clustering of the correlation matrix, which is a static tree-based method. The authors of this paper compare the performance of the standard HRP with other static and also adaptive tree-based methods, but also seriation-based methods that do not rely on trees. Seriation is a broader concept allowing to reorder the rows or columns of a matrix to best express similarities between the elements. Each discussed variation leads to a different time series reflecting portfolio performance using a 20-year backtest of a multi-asset futures universe. An unsupervised representation learning based on this time series data creates a taxonomy that groups the strategies in high correspondence to the structure of the various types of ASRP. The performance analysis of the variations shows that most of the static tree-based alternatives of HRP outperform the single linkage clustering used in HRP on a risk-adjusted basis. Adaptive tree methods show mixed results and most generic seriation-based approaches underperform.

Performance Evaluation of the Data Clustering Techniques and Cluster Validity Indices for Efficient Toolpath Development for Incremental Sheet Forming

Abstract The goal of current research is to compare the data clustering techniques and cluster validity indices for geometrical feature extraction using point cloud. Here, the point clouds are generated by slicing of the computer-aided design (CAD) surface, and the data on each slice is used as inputs to the clustering algorithms. The clustering techniques are used to detect the multiple closed contours on the planer datasets. In this paper, the four most popular clustering techniques, i.e., partition-based (K-means), density-based (DBSCAN), single linkage hierarchical clustering, a variant of hierarchical clustering, and graph-based (spectral) clustering technique are compared using four different datasets. The single linkage hierarchical clustering is preferred over the other variants as it detects the arbitrarily shaped clusters efficiently and effectively. The comparison is based on the ability of successful detection of the closed contours on the planer dataset, the time required, and the input needed for the algorithm. From the investigations, it is found that DBSCAN is the most suitable technique for the feature-based toolpaths (FBTs) development. Besides, for the quality assessment of the clustering solutions and to pinpoint the superlative validity indices, techniques like Calinski-Harabasz, Ball-Hall, Davies-Bouldin, Dunn, Det Ratio, Silhouette, Trace WiB, and Log Det Ratio are compared. The solutions of the clustering techniques are used to compare the validity indices. On the basis of comparative analysis, it is concluded that the Ball-Hall, Dunn, Det Ratio, and Log Det Ratio indices are the best validity criterions for the arbitrary shaped closed contours. The overall outcomes of this research will help in building the algorithms for the feature-based toolpath development strategies for the various manufacturing processes using data science and machine learning techniques.

Maturational delay and asymmetric information flow of brain connectivity in SHR model of ADHD revealed by topological analysis of metabolic networks

Attention-deficit hyperactivity disorder (ADHD) is a complex brain development disorder characterized by hyperactivity/impulsivity and inattention. A major hypothesis of ADHD is a lag of maturation, which is supported mainly by anatomical studies evaluating cortical thickness. Here, we analyzed changes of topological characteristics of whole-brain metabolic connectivity in twelve SHR rats selected as ADHD-model rats by confirming behavior abnormalities using the marble burying test, open field test, and delay discounting task and 12 Wistar Kyoto rats as the control group, across development from 4 weeks old (childhood) and 6 weeks old (entry of puberty). A topological approach based on graph filtrations revealed a lag in the strengthening of limbic-cortical/subcortical connections in ADHD-model rats. This in turn related to impaired modularization of memory and reward-motivation associated regions. Using mathematical network analysis techniques such as single linkage hierarchical clustering and volume entropy, we observed left-lateralized connectivity in the ADHD-model rats at 6 weeks old. Our findings supported the maturational delay of metabolic connectivity in the SHR model of ADHD, and also suggested the possibility of impaired and compensative reconfiguration of information flow over the brain network.

Chain-detection Between Clusters

Chains connecting two or more different clusters are a well known problem of clustering algorithms like DBSCAN or Single Linkage Clustering. Since already a small number of points resulting from, e. g., noise can form such a chain and build a bridge between different clusters, it can happen that the results of the clustering algorithm are distorted: several disparate clusters get merged into one. This single-link effect is rather known but to the best of our knowledge there are no satisfying solutions which extract those chains, yet. We present a new algorithm detecting not only straight chains between clusters, but also bent and noisy ones. Users are able to choose between eliminating one dimensional and higher dimensional chains connecting clusters to receive the underlying cluster structure. Also, the desired straightness can be set by the user. As this paper is an extension of [8], we apply our technique not only in combination with DBSCAN but also with single link hierarchical clustering. On a real world dataset containing traffic accidents in Great Britain we were able to detect chains emerging from streets between cities and villages, which led to clusters composed of diverse villages. Additionally, we analyzed the robustness regarding the variance of chains in synthetic experiments.

TriplClust: An Algorithm for Curve Detection in 3D Point Clouds

In this article, we describe an algorithm for detecting and separating curves in 3D point clouds without making a priori assumptions about their parametric shape. The algorithm is called 'TriplClust' because it is based on the idea of clustering point triplets instead of the original points. We define a distance measure on point triplets and then apply a single-link hierarchical clustering on the triplets. The clustering process can be controlled by several parameters, which are described in detail, and suggestions for reasonable choices for these parameters based on the input data are made. Moreover, we suggest a simple criterion for stopping the single link clustering automatically.

Automatic trajectory recognition in Active Target Time Projection Chambers data by means of hierarchical clustering

The automatic reconstruction of three-dimensional particle tracks from Active Target Time Projection Chambers data can be a challenging task, especially in the presence of noise. In this article, we propose a non-parametric algorithm that is based on the idea of clustering point triplets instead of the original points. We define an appropriate distance measure on point triplets and then apply a single-link hierarchical clustering on the triplets. Compared to parametric approaches like RANSAC or the Hough transform, the new algorithm has the advantage of potentially finding trajectories even of shapes that are not known beforehand. This feature is particularly important in low-energy nuclear physics experiments with Active Targets operating inside a magnetic field. The algorithm has been validated using data from experiments performed with the Active Target Time Projection Chamber developed at the National Superconducting Cyclotron Laboratory (NSCL). The results demonstrate the capability of the algorithm to identify and isolate particle tracks that describe non-analytical trajectories. For curved tracks, the vertex detection recall was 86% and the precision 94%. For straight tracks, the vertex detection recall was 96% and the precision 98%. In the case of a test set containing only straight linear tracks, the algorithm performed better than an iterative Hough transform.

Optimized Threshold Inference for Partitioning of Clones From High-Throughput B Cell Repertoire Sequencing Data.

During adaptive immune responses, activated B cells expand and undergo somatic hypermutation of their B cell receptor (BCR), forming a clone of diversified cells that can be related back to a common ancestor. Identification of B cell clones from high-throughput Adaptive Immune Receptor Repertoire sequencing (AIRR-seq) data relies on computational analysis. Recently, we proposed an automated method to partition sequences into clonal groups based on single-linkage hierarchical clustering of the BCR junction region with length-normalized Hamming distance metric. This method could identify clonal sequences with high confidence on several benchmark experimental and simulated data sets. However, determining the threshold to cut the hierarchy, a key step in the method, is computationally expensive for large-scale repertoire sequencing data sets. Moreover, the methodology was unable to provide estimates of accuracy for new data. Here, a new method is presented that addresses this computational bottleneck and also provides a study-specific estimation of performance, including sensitivity and specificity. The method uses a finite mixture model fitting procedure for learning the parameters of two univariate curves which fit the bimodal distribution of the distance vector between pairs of sequences. These distributions are used to estimate the performance of different threshold choices for partitioning sequences into clones. These performance estimates are validated using simulated and experimental data sets. With this method, clones can be identified from AIRR-seq data with sensitivity and specificity profiles that are user-defined based on the overall goals of the study.

True-link clustering through signaling process and subcommunity merge in overlapping community detection

To overcome the difficulty in detecting reliable overlapping communities in complex networks, “true-link” and “pseudo-link” are firstly proposed on the basis of the original network graph. Then, the “true-link” graph is obtained through the preprocessing of the original network graph. And then the line graph is partitioned by means of signaling process and single-linkage hierarchical clustering. Meanwhile, the subcommunities are merged based on the proposed similarity between communities, which eradicates the inherently redundant overlapping communities to a certain extent. Compared with other overlapping community detection algorithms, this proposed algorithm is of strong robustness and high accuracy. All the results of the experiments boil down to the conclusion that this True-link Clustering Community Detection is an overlapping community detection algorithm prevailing over others.

Statistical properties of the single linkage hierarchical clustering estimator

Distance-based hierarchical clustering (HC) methods are widely used in unsupervised data analysis but few authors take account of uncertainty in the distance data. We incorporate a statistical model of the uncertainty through corruption or noise in the pairwise distances and investigate the problem of estimating the HC as unknown parameters from measurements. Specifically, we focus on single linkage hierarchical clustering (SLHC) and study its geometry. We prove that under fairly reasonable conditions on the probability distribution governing measurements, SLHC is equivalent to maximum partial profile likelihood estimation (MPPLE) with some of the information contained in the data ignored. At the same time, we show that direct evaluation of SLHC on maximum likelihood estimation (MLE) of pairwise distances yields a consistent estimator. Consequently, a full MLE is expected to perform better than SLHC in getting the correct HC results for the ground truth metric.

Peringkasan Tweet Berdasarkan Trending Topic Twitter Dengan Pembobotan TF-IDF dan Single Linkage AngglomerativeHierarchical Clustering

Trending topic is a feature provided by twitter that informs something widely discussed by users in a particular time. The form of a trending topic is a hashtag and can be selected by clicking. However, the number of tweets for each trending topics can be very large, so it will be difficult if we want to know all the contents. So, in order to make easy when reading the topic, a small number of tweets can be selected as the main idea of the topic. In this study, we applied the Agglomerative Single Linkage Hierarchical Clustering by calculating the TF-IDF value for each word in advance. We used 100 trending topics, where each topic consists of 50 tweets in Indonesian. For testing, we provided 30 trending topics which consist of 2 until 9 sub-topics. The result is that each trending topics can be summarized into shorter text contains 2 until 9 tweets. We were able to summarize 1 trending topics exactly same as the topic summarized by human expert. However, the rest of topics corresponded partially with human expert.

Asymptomatic Plasmodium falciparum infection in children is associated with increased auto-antibody production, high IL-10 plasma levels and antibodies to merozoite surface protein 3.

BackgroundMechanisms of acquired protection to malaria in asymptomatic Plasmodium falciparum carriers are only partially understood. Among them, the role plays by the self-reactive antibodies has not been clarified yet. In this study, the relationship between repertoires of circulating self-reactive and parasite-specific immunoglobulin G (IgG), their correlation with cytokine levels, and their association with protection against malaria was investigated in asymptomatic Plasmodium falciparum-infected Gabonese children.MethodsThe diversity of P. falciparum-specific antibody repertoire was analysed using a protein micro-array immunoassay, the total auto-antibody repertoire by quantitative immunoblotting and circulating cytokine levels were measured by ELISA in endemic controls (EC) and P. falciparum-infected children from Gabon with asymptomatic (AM) or mild malaria (MM). The association of self- and parasite-specific antibody repertoires with circulating cytokines was evaluated using single linkage hierarchical clustering, Kruskal – Wallis tests and Spearman’s rank correlation.ResultsChildren with AM exhibited an IgG response to merozoite surface protein 3 (MSP3) but not to MSP1-19, although their levels of total P. falciparum-specific IgG were similar to those in the MM group. Moreover, the asymptomatic children had increased levels of autoantibodies recognising brain antigens. In addition, a correlation between IL-10 levels and parasite load was found in AM and MM children. These two groups also exhibited significant correlations between plasma levels of IL-10 and IFN-γ with age and with total plasma IgG levels. IL-10 and IFN-γ levels were also associated with auto-antibody responses in AM.ConclusionsAltogether, these results indicate that a self-reactive polyclonal response associated with increased IgG to MSP3 and high plasma levels of IL-10 and IFN-γ may contribute to protective immune mechanisms triggered in asymptomatic P. falciparum infection in Gabonese children.Electronic supplementary materialThe online version of this article (doi:10.1186/s12936-015-0658-7) contains supplementary material, which is available to authorized users.

On clustering procedures and nonparametric mixture estimation

This paper deals with nonparametric estimation of conditional densities in mixture models in the case when additional covariates are available. The proposed approach consists of performing a preliminary clustering algorithm on the additional covariates to guess the mixture component of each observation. Conditional densities of the mixture model are then estimated using kernel density estimates applied separately to each cluster. We investigate the expected $L_{1}$-error of the resulting estimates and derive optimal rates of convergence over classical nonparametric density classes provided the clustering method is accurate. Performances of clustering algorithms are measured by the maximal misclassification error. We obtain upper bounds of this quantity for a single linkage hierarchical clustering algorithm. Lastly, applications of the proposed method to mixture models involving electricity distribution data and simulated data are presented.

Statistical properties of convex clustering.

In this manuscript, we study the statistical properties of convex clustering. We establish that convex clustering is closely related to single linkage hierarchical clustering and k-means clustering. In addition, we derive the range of the tuning parameter for convex clustering that yields a non-trivial solution. We also provide an unbiased estimator of the degrees of freedom, and provide a finite sample bound for the prediction error for convex clustering. We compare convex clustering to some traditional clustering methods in simulation studies.

The cluster graphical lasso for improved estimation of Gaussian graphical models

The task of estimating a Gaussian graphical model in the high-dimensional setting is considered. The graphical lasso, which involves maximizing the Gaussian log likelihood subject to a lasso penalty, is a well-studied approach for this task. A surprising connection between the graphical lasso and hierarchical clustering is introduced: the graphical lasso in effect performs a two-step procedure, in which (1) single linkage hierarchical clustering is performed on the variables in order to identify connected components, and then (2) a penalized log likelihood is maximized on the subset of variables within each connected component. Thus, the graphical lasso determines the connected components of the estimated network via single linkage clustering. The single linkage clustering is known to perform poorly in certain finite-sample settings. Therefore, the cluster graphical lasso, which involves clustering the features using an alternative to single linkage clustering, and then performing the graphical lasso on the subset of variables within each cluster, is proposed. Model selection consistency for this technique is established, and its improved performance relative to the graphical lasso is demonstrated in a simulation study, as well as in applications to a university webpage and a gene expression data sets.