Advanced Clustering Techniques Research Articles

High-Dimensional Fluctuations in Liquid Water: Combining Chemical Intuition with Unsupervised Learning

The microscopic description of the local structure of water remains an open challenge. Here, we adopt an agnostic approach to understanding water's hydrogen bond network using data harvested from molecular dynamics simulations of an empirical water model. A battery of state-of-the-art unsupervised data-science techniques are used to characterize the free-energy landscape of water starting from encoding the water environment using local atomic descriptors, through dimensionality reduction and finally the use of advanced clustering techniques. Analysis of the free energy under ambient conditions was found to be consistent with a rough single basin and independent of the choice of the water model. We find that the fluctuations of the water network occur in a high-dimensional space, which we characterize using a combination of both atomic descriptors and chemical-intuition-based coordinates. We demonstrate that a combination of both types of variables is needed in order to adequately capture the complexity of the fluctuations in the hydrogen bond network at different length scales both at room temperature and also close to the critical point of water. Our results provide a general framework for examining fluctuations in water under different conditions.

ActivityExplorer: A semi-supervised approach to discover unknown activity classes in HAR systems

• A semi-supervised activity recognition approach able to identify unknown activity classes. • Approach designed to deal with small-sample set scenarios with limited amount of training data. • Approach based on affinity propagation clustering able to automatically identify the number of clusters. • Results on public datasets confirm the efficacy of the proposed approach. A semi-supervised activity recognition system is here proposed to deal with partially labeled video-sequences, where the uncertainty in the data comes from two different factors: only a subset of the data has a class label assigned and only part of the activity classes are known. In particular, the paper presents ActivityExplorer, an approach able to identify clusters of similar activity patterns within the dataset and to identify those clusters that might correspond to new activity classes, still unknown to the recognition system. These capabilities are realized thanks to a combination of metric learning, used to determine a suitable subspace for pattern classification, an advanced clustering technique and ad hoc indicators defined to estimate the membership of each pattern to known classes and possibly identify new activities.

The (im)possibility of strong chemical tagging

Context.The possibility of identifying co-natal stars that have dispersed into the Galactic disc based on chemistry alone is called strong chemical tagging. It has been debated for a long time whether this is indeed feasible; it holds the promise of reconstructing the detailed star formation history of a large fraction of stars in the Galactic disc.Aims.We investigate the feasibility of strong chemical tagging using known member stars of open clusters.Methods.We analysed the largest sample of cluster members that have been homogeneously characterised with high-resolution differential abundances for 16 different elements. We also investigated the possibility of finding the known clusters in the APOGEE DR16 red clump sample with 18 chemical species. For both purposes, we used a clustering algorithm and an unsupervised dimensionality reduction technique to blindly search for groups of stars in chemical space.Results.Even if the internal coherence of the stellar abundances in the same cluster is high, typically 0.03 dex, the overlap in the chemical signatures of the clusters is large. In the sample with the highest precision and no field stars, we only recover 9 out of the 31 analysed clusters at a 40% threshold of homogeneity and precision. This ratio slightly increases when we only use clusters with 7 or more members. In the APOGEE sample, field stars are present along with four populated clusters. In this case, only one of the open clusters was moderately recovered.Conclusions.In our best-case scenario, more than 70% of the groups of stars are in fact statistical groups that contain stars belonging to different real clusters. This indicates that the chances of recovering the majority of birth clusters dissolved in the field are slim, even with the most advanced clustering techniques. We show that different stellar birth sites can have overlapping chemical signatures, even when high-resolution abundances of many different nucleosynthesis channels are used. This is substantial evidence against the possibility of strong chemical tagging. However, we can hope to recover some particular birth clusters that stand out at the edges of the chemical distribution.

P-028. Study of mean platelet volume in first trimester of pregnancy and its association with development of preeclampsia

A semi-supervised activity recognition system is here proposed to deal with partially labeled video-sequences, where the uncertainty in the data comes from two different factors: only a subset of the data has a class label assigned and only part of the activity classes are known. In particular, the paper presents ActivityExplorer, an approach able to identify clusters of similar activity patterns within the dataset and to identify those clusters that might correspond to new activity classes, still unknown to the recognition system. These capabilities are realized thanks to a combination of metric learning, used to determine a suitable subspace for pattern classification, an advanced clustering technique and ad hoc indicators defined to estimate the membership of each pattern to known classes and possibly identify new activities.

유동 클러스터 탐지 기법의 개발과 적용: 서울시 대중교통 통행을 사례로

Smart-card data is a high quality data that can be used in analyzing urban flow patterns. However, due to methodological difficulties of flow clustering, earlier studies have not been able to take full advantage of the smart-card data. To overcome this limitation, this study introduces an efficient flow cluster detection method based on smart-card data of public transportation and applies it to public transit trips in Seoul to derive significant flow clusters of trips. As a result of case study, we identified various clusters of public trasit trips in Seoul and could detect significant clusters that are composed of several individual travel flows being adjacent in space but mostly out of different transport modes and routes. Furthermore, the methodology of this study made it possible to capture the major travel flow patterns. They have not been recognized as important flow patterns in a normal way, but they have significant traffic volume when gathered into a flow cluster. This study provides important methodological implications to the analysis of flow patterns, and the suggested algorithm is expected to be the basis of more advanced flow clustering techniques in the future.

An Insurtech for the Online Insurance: A Customer Repurchasing Behavior Study in Taiwan

This study explores online insurance ventures by customers’ behavioral profiles from data mining and visualization. First, we employ a simple decision-tree statistical learning method on unique hand-collected and processed data in Taiwan. This method considers first-time online insurance subscribers as new ones without a proper marketing connection after six months. Second, using advanced clustering technique and decision tree statistical learning on first-time purchasing customers, we find they repurchase online travel insurance for different purposes varying periods. Finally, we get robust results engaging different segmentations of customer data. In these ways, we enable marketing strategies to work with generating decision rules.

Identification of low energy nuclear recoils in a gas time projection chamber with optical readout

The search for a novel technology, which is able to detect and reconstruct nuclear recoil events in the keV energy range, has become increasingly important now that vast regions of high mass weakly-interacting-massive-particle–like dark matter candidates have been excluded. Gaseous time projection chambers (TPC) with optical readout are very promising candidates combining the complete event information provided by the TPC technique with the high sensitivity and granularity of the latest generation light sensors. A TPC with an amplification at the anode, obtained with gas electron multipliers (GEMs), was tested at the Laboratori Nazionali di Frascati. Photons and neutrons from radioactive sources were employed to induce recoiling nuclei and electrons with kinetic energy in the range 1–100 keV. A He-CF4 (60/40) gas mixture was used at atmospheric pressure and the light produced during the multiplication in the GEM channels was acquired by a high-position resolution and low-noise complementary metal-oxide semiconductor camera and a photomultiplier. A multi-stage pattern recognition algorithm based on an advanced clustering technique is presented here. A number of cluster-shaped observables are used to identify nuclear recoils induced by neutrons, which originated from a AmBe source against x-ray 55Fe photoelectrons. An efficiency of 18% to detect nuclear recoils with an energy of about 6 keV is reached, while suppressing 96% of the 55Fe photoelectrons, making this optical read-out gas TPC a very promising candidate for future investigations of ultra-rare events such as directional direct dark matter searches.

Improved Density Peaks Clustering Based on Natural Neighbor Expanded Group

Density peaks clustering (DPC) is an advanced clustering technique due to its multiple advantages of efficiently determining cluster centers, fewer arguments, no iterations, no border noise, etc. However, it does suffer from the following defects: (1) difficult to determine a suitable value of its crucial cutoff distance parameter, (2) the local density metric is too simple to find out the proper center(s) of the sparse cluster(s), and (3) it is not robust that parts of prominent density peaks are remotely assigned. This paper proposes improved density peaks clustering based on natural neighbor expanded group (DPC-NNEG). The cores of the proposed algorithm contain two parts: (1) define natural neighbor expanded (NNE) and natural neighbor expanded group (NNEG) and (2) divide all NNEGs into a goal number of sets as the final clustering result, according to the closeness degree of NNEGs. At the same time, the paper provides the measurement of the closeness degree. We compared the state of the art with our proposal in public datasets, including several complex and real datasets. Experiments show the effectiveness and robustness of the proposed algorithm.

Corporate image or social engagement: Twitter discourse on corporate social responsibility (CSR) in public relations strategies in the energy sector

Social media have opened up new opportunities for the creation of innovative public relations strategies focused on establishing and cultivating relationships with stakeholders on the basis of meaningful dialogue. Consideration of the interrelation between corporate social responsibility (CSR) and public relations highlights new areas for exploration and engagement. Both the dialogical and semantic perspectives reveal the performative and conversational aspects of social media. In general, both the linguistic panorama of CSR and digital media as part of a PR strategy open new possibilities for a dialogical, interactive, meaningful relationship strategy for corporate image management. Based on the linguistic approach to CSR and the Communication Management Approach, this paper explores the linguistic use of Twitter as a primary dialogical strategy to effectively enhance interactive dialogue-based relationships with the stakeholders of the top 50 companies in the energy sector based on tweet data from 2016. Semantic analysis was conducted by advanced text mining and clustering techniques on 3042 tweets monitored in 2017 that contained the leading CSR-related hashtags and keywords. The results demonstrated that the top energy companies apply a defensive and symbolic perspective, mainly for branding purposes. The corporate discourse dominates over a meaningful conversational strategy to foster interaction with stakeholders around sustainability issues on Twitter. The study reveals a homogenized interrelation between CSR, social media, and public relations. The results reveal a tendency for isomorphy in the communication models applied by the companies in the energy sector. Furthermore, similarities in semantics and thus strong tendencies to mutually mimic dialogical strategies are also observed. The semantic narrative built around the brand indicates a limited orientation towards CSR and sustainability. As such, it does not contribute to the creation of a dialogical interaction and meaningful relationships with multiple stakeholders on Twitter, in the high-risk sector represented by the energy industry.

Advanced Clustering Technique to Handle Multi-word Expressions for Descriptive Documents

Expressive clustering comprises of naturally sorting out information occurrences into groups and creating a graphic outline for each group. The portrayal ought to advise a client about the substance regarding each group moving forward without any more assessment of the particular occasions, empowering a client to quickly filter for pertinent bunches. Choice of portrayals frequently depends on heuristic criteria. We model graphic grouping as an auto-encoder organize that predicts highlights from bunch assignments and predicts bunch assignments from a subset of highlights. We present an area free bunching based methodology for programmed extraction of multiword expressions (MWEs). The strategy consolidates factual data from a universally useful corpus and writings from Wikipedia articles. We fuse affiliation measures through elements of information focuses to bunch MWEs and after that process the positioning score for each MWE dependent on the nearest model doled out to a group. Assessment results, accomplished for two dialects, demonstrate that a mix of affiliation estimates gives an improvement in the positioning of MWEs contrasted and basic checks of co event frequencies and simply factual measures.

Lung Lesion Classification Scheme Using Optimization Techniques and Hybrid (KNN-SVM) Classifier

This paper proposes an intellectual classification system to classify normal and abnormal CT scan lung images. Prediction of cancer cells from lung image is such a difficult task for physicians and researchers. A novel technique for lung tumour detection with Grey Wolf Optimized & Whale Optimization Algorithm – Support Vector Machine (GWO & WOA-SVM) is proposed. CT scan images are used as input images. Investigations focused on the segmentation and classification part to find the lung lesion region. For finding the lung abnormality and lesion region in the lung, four different stages are used. The first stage is the image acquisition process. The second stage is the image pre-processing and enhancement stage, here Advanced Clustering (AC) technique is used. The next stage is the segmentation process, here the advanced surface normal overlap (ASNO) lung segmentation algorithm is used. Finally, the lung lesion is classified using the hybrid classifier followed by a different optimization technique. The research gap for this paper is to find the lung lesion using an optimization-based hybrid classification technique. Nowadays, classification decision and treatment of lung tumours are based on symptoms and radiological appearance. Here Hybrid Classifier (KNN-SVM) is used to classify 700 images; it is observed from the results that the Hybrid classifier KNN-SVM demonstrated the highest classification accuracy rate of 97.6% among other methods.

Implementation of Software Agents and Advanced AoA for Disease Data Analysis.

To eliminate the possibilities of getting various contradicting solutions to a single problem during diagnosis, a single regular Agent oriented Approach (AoA) is replaced by Intelligent Artificial Agents that act like human and even dynamically decide in any situations known as Intelligent Searching Approach (ISA) is proposed. These agents are used to analyse the medical forums and results or findings are derived accurately than any manual approach. Multiple Agents have been used to analyse the blogs by dividing the work areas and communicating themselves using Agent Communication Language (ACL) and FIPA. The local solutions thus formed are forwarded to a global agent. This Global Agent controls all operations and makes the decision about the best solution. As the Global Agent controls all other agents, it eradicates unwanted and ineffective communication between the various local agents and hence keeping the time taken for communication at the minimum level. Based on these solutions a prioritization matrix is formed using advanced clustering techniques to create a prioritized content of suggested best solutions. Once the decision is made, the refining process runs several times recursively checking for all possible better solutions solving the input. On completion of this process, the Global Agent returns the exact result of the discussion. This process saves time rather than researching the entire blog for result data. This advanced approach lights a different way of obtaining solution keeping the time taken for discussion and intercommunication between the agents to the minimal level but not compromising on the perfection of the solution at the same time.

The mechanism of RNA base fraying: Molecular dynamics simulations analyzed with core-set Markov state models.

The process of RNA base fraying (i.e., the transient opening of the termini of a helix) is involved in many aspects of RNA dynamics. We here use molecular dynamics simulations and Markov state models to characterize the kinetics of RNA fraying and its sequence and direction dependence. In particular, we first introduce a method for determining biomolecular dynamics employing core-set Markov state models constructed using an advanced clustering technique. The method is validated on previously reported simulations. We then use the method to analyze extensive trajectories for four different RNA model duplexes. Results obtained using D. E. Shaw research and AMBER force fields are compared and discussed in detail and show a non-trivial interplay between the stability of intermediate states and the overall fraying kinetics.

Clustering by reordering of similarity and Laplacian matrices: Application to galaxy clusters

Abstract Similarity metrics, kernels and similarity-based algorithms have gained much attention due to their increasing applications in information retrieval, data mining, pattern recognition and machine learning. Similarity Graphs are often adopted as the underlying representation of similarity matrices and are at the origin of known clustering algorithms such as spectral clustering. Similarity matrices offer the advantage of working in object–object (two-dimensional) space where visualization of clusters similarities is available instead of object-features (multi-dimensional) space. In this paper, sparse ϵ -similarity graphs are constructed and decomposed into strong components using appropriate methods such as Dulmage–Mendelsohn permutation (DMperm) and/or Reverse Cuthill–McKee (RCM) algorithms. The obtained strong components correspond to groups (clusters) in the input (feature) space. Parameter ϵ i is estimated locally, at each data point i from a corresponding narrow range of the number of nearest neighbors. Although more advanced clustering techniques are available, our method has the advantages of simplicity, better complexity and direct visualization of the clusters similarities in a two-dimensional space. Also, no prior information about the number of clusters is needed. We conducted our experiments on two and three dimensional, low and high-sized synthetic datasets as well as on an astronomical real-dataset. The results are verified graphically and analyzed using gap statistics over a range of neighbors to verify the robustness of the algorithm and the stability of the results. Combining the proposed algorithm with gap statistics provides a promising tool for solving clustering problems. An astronomical application is conducted for confirming the existence of 45 galaxy clusters around the X-ray positions of galaxy clusters in the redshift range [0.1..0.8]. We re-estimate the photometric redshifts of the identified galaxy clusters and obtain acceptable values compared to published spectroscopic redshifts with a 0.029 standard deviation of their differences.

A methodology to discover and understand complex patterns: Interpreted Integrative Multiview Clustering (I2MC)

The main goal of this work is to develop a methodology for finding nutritional patterns from a variety of individual characteristics which can contribute to better understand the interactions between nutrition and health, provided that the complexity of the phenomenon gives poor performance using classical approaches. An innovative methodology based on a combination of advanced clustering techniques and consistent conceptual interpretation of clusters is proposed to find more understandable patterns or clusters. The Interpreted Integrative Multiview Clustering (I2MC) combines the previously proposed Integrative Multiview Clustering (IMC) with a new interpretation methodology NCIMS. IMC uses crossing operations over the several partitions obtained with the different views. Comparison with other classical clustering techniques is provided to assess the performance of this approach. IMC helps to reduce the high dimensionality of the data based on multiview division of variables.Two innovative Cluster Interpretation methodologies are proposed to support the understanding of the clusters. These are automatic methods to detect the significant variables that describe the clusters; also, a mechanism to deal with the consistency between the interpretations inter clusters of a single partition CI-IMS, or between pairs of nested partitions NCIMS. Some formal concepts are specifically introduced to be used in the NCIMS.I2MC is used to validate the interpretability of the participant’s profiles from an intervention nutritional study. The method has advantages to deal with complex datasets including heterogeneous variables corresponding to different topics and is able to provide meaningful partitions.

A highly scalable clustering scheme using boundary information

Many advanced clustering techniques are effective in dealing datasets in complicated situations. However, when facing large datasets, which are increasingly common in the era of big data, the time requirements of most existing techniques can quickly become intolerable. To tackle this challenge, in this paper, we propose Scalable Clustering Using Boundary Information (SCUBI), a highly flexible and scalable clustering scheme. The idea of SCUBI is to identify the boundary points of the original dataset in the first place and then group boundary points into suitable clusters using existing clustering techniques. Finally, the rest points are assigned to the same cluster as their nearest boundary points. To demonstrate the effectiveness and scalability of SCUBI, we plug the well-known DBSCAN algorithm into SCUBI. Comprehensive experiments are conducted using datasets with up to two million data points to compare the clustering results and time efficiency between DBSCAN and SCUBI-DBSCAN. Experimental results show that our method can obtain almost identical clustering results as the standard DBSCAN while achieving orders of magnitude speedup especially on large datasets, which confirms the scalability of SCUBI. Experiments are also performed on other clustering algorithms with high time complexity to verify the flexibility of SCUBI.

Single tree identification using airborne multibaseline SAR interferometry data

Remote sensing data allowlarge scale observation of forested ecosystems. Forest assessment benefits from information about individual trees. Multibaseline SAR interferometry (InSAR) is able to generate dense point clouds of forest canopies, similar to airborne laser scanning (ALS). This type of point cloud was generated using data from the Ka-bandMEMPHIS system, acquired over a mainly coniferous forest near Vordemwald in the Swiss Midlands. This point cloud was segmented using an advanced clustering technique to detect individual trees and derive their positions, heights, and crown diameters. To evaluate the InSAR point cloud properties and limitations, it was compared to products derived from ALS and stereo-photogrammetry. All point clouds showed similar geolocation accuracies with 0.2–0.3mrelative shifts. Both InSAR and photogrammetry techniques yielded points predominantly located in the upper levels of the forest vegetation, while ALS provided points fromthe top of the canopy down to the understory and forest floor. The canopy height models agreed very well with each other, with R² values between 0.84 and 0.89. The detected trees and their estimated physical and structural parameters were validated by comparing them to reference forestry data. A detection rate of ~90% was achieved for larger trees, corresponding to half of the reference trees. The smaller trees were detected with a success rate of ~50%. The tree height was slightly underestimated, with a R² value of 0.63. The estimated crown diameter agreed on an average sense, however with a relatively low R² value of 0.19. Very high success rates (N90%) were obtained when matching the trees detected fromthe InSAR-datawith those detected fromthe ALS- and photogrammetry-data. There, InSAR tree heightswere in themean 1–1.5mlower,with high R² values ranging between 0.8 and 0.9. Our results demonstrate the use of millimeter wave SAR interferometry data as an alternative to ALS- and photogrammetry-based data for forest monitoring.

Sampling Molecular Conformers in Solution with Quantum Mechanical Accuracy at a Nearly Molecular-Mechanics Cost.

We introduce a method to evaluate the relative populations of different conformers of molecular species in solution, aiming at quantum mechanical accuracy, while keeping the computational cost at a nearly molecular-mechanics level. This goal is achieved by combining long classical molecular-dynamics simulations to sample the free-energy landscape of the system, advanced clustering techniques to identify the most relevant conformers, and thermodynamic perturbation theory to correct the resulting populations, using quantum-mechanical energies from density functional theory. A quantitative criterion for assessing the accuracy thus achieved is proposed. The resulting methodology is demonstrated in the specific case of cyanin (cyanidin-3-glucoside) in water solution.

Hybrid positioning data fusion in heterogeneous networks with critical hearability

In this paper, we propose and investigate a hybrid positioning data fusion technique for heterogeneous networks in critical transmission scenarios. The focus is on two scenarios: the small indoor scenario combining Wi-Fi and cellular systems and the small-to-mid-scale scenario composed of one located Mobile Terminal (MT) and one anchor node (AN). More specifically, we investigate the effect of the availability of three metrics i.e. the time of arrival (ToA), the angle of arrival (AoA), and the received signal strength-based fingerprint (RSS) on the positioning accuracy when the number of ANs is less than three. To combine these measurements, we use a 2-level unscented Kalman Filter (UKF) in conjunction with some advanced clustering techniques based on genetic algorithms. Simulation results show that the proposed hybrid data fusion technique outperforms the techniques presented in the literature independently of the transmission conditions.

Energy and Water Consumption Variability in School Buildings: Review and Application of Clustering Techniques

AbstractIn developed countries, the building sector is responsible for a very significant share of the total energy consumption. School buildings, because they are places where children are educated and learn to become active members of the society, should be a good example of an efficient use of energy and water. In this study, data of the energy and water consumption of 23 Portuguese schools and their main building characteristics and properties were gathered. This information was normalized to homogenize the data set and then analyzed using advanced clustering techniques. The results show a significant variability in the consumption of different schools, even with similar characteristics, suggesting that the user behavior plays an important role in their efficiency. The complete linkage and Ward’s clustering methods were applied, both produced three clusters, and reference values for electricity and water consumption were defined.