Multidimensional Visualization Technique Research Articles

Kurdish Dialect Recognition using 1D CNN

Dialect recognition is one of the most attentive topics in the speech analysis area. Machine learning algorithms have been widely used to identify dialects. In this paper, a model that based on three different 1D Convolutional Neural Network (CNN) structures is developed for Kurdish dialect recognition. This model is evaluated, and CNN structures are compared to each other. The result shows that the proposed model has outperformed the state of the art. The model is evaluated on the experimental data that have been collected by the staff of department of computer science at the University of Halabja. Three dialects are involved in the dataset as the Kurdish language consists of three major dialects, namely Northern Kurdish (Badini variant), Central Kurdish (Sorani variant), and Hawrami. The advantage of the CNN model is not required to concern handcraft as the CNN model is featureless. According to the results, the 1 D CNN method can make predictions with an average accuracy of 95.53% on the Kurdish dialect classification. In this study, a new method is proposed to interpret the closeness of the Kurdish dialects by using a confusion matrix and a non-metric multi-dimensional visualization technique. The outcome demonstrates that it is straightforward to cluster given Kurdish dialects and linearly isolated from the neighboring dialects.

Usability Assessment of the Visualization-Driven Approaches to the HVAC Data Exploration

Application of the Internet-connected operational devices in the heating, ventilation and conditioning (HVAC) systems has extended the cyber-attack surface by introducing different malicious scenarios. The analysis of the HVAC data may provide insight on typical patterns of the system operations. Implementation of the thoroughly elaborated visualization models may significantly increase the efficiency of the suspicious activity identification in the HVAC systems. In the paper we present the results of the laboratory usability testing of three visualization models used to analyze HVAC data – matrix-based visualization technique, non-linear multidimensional visualization technique RadViz and timeline chart. Matrix-based visualization and RadViz visualization are often used in anomaly detection process, while timeline charts are a traditional way to present operational HVAC data. We describe the experiment design and discuss the results obtained. The usability testing revealed advantages and limitations of these visualization techniques in behavior pattern and anomaly identification tasks. The results can further serve as guidelines for task-dependent selection of a visualization technique.

Painting image browser applying an associate-rule-aware multidimensional data visualization technique

Exploration of artworks is enjoyable but often time consuming. For example, it is not always easy to discover the favorite types of unknown painting works. It is not also always easy to explore unpopular painting works which looks similar to painting works created by famous artists. This paper presents a painting image browser which assists the explorative discovery of user-interested painting works. The presented browser applies a new multidimensional data visualization technique that highlights particular ranges of particular numeric values based on association rules to suggest cues to find favorite painting images. This study assumes a large number of painting images are provided where categorical information (e.g., names of artists, created year) is assigned to the images. The presented system firstly calculates the feature values of the images as a preprocessing step. Then the browser visualizes the multidimensional feature values as a heatmap and highlights association rules discovered from the relationships between the feature values and categorical information. This mechanism enables users to explore favorite painting images or painting images that look similar to famous painting works. Our case study and user evaluation demonstrates the effectiveness of the presented image browser.

Kurdish Dialects and Neighbor Languages Automatic Recognition

Dialect recognition is one of the most hot topics in the speech analysis area. In this study a system for dialect and language recognition is developed using phonetic and a style based features. The study suggests a new set of feature using one-dimensional LBP feature. The results show that the proposed LBP set of feature is useful to improve dialect and language recognition accuracy. The acquired data involved in this study are three Kurdish dialects (Sorani, Badini and Hawrami) with three neighbor languages (Arabic, Persian and Turkish). The study proposed a new method to interpret the closeness of the Kurdish dialects and their neighbor languages using confusion matrix and a non-metric multi-dimensional visualization technique. The result shows that the Kurdish dialects can be clustered and linearly separated from the neighbor languages.

A data-driven multidimensional visualization technique for process fault detection and diagnosis

A multidimensional visualization technique is described for fault detection and diagnosis of a multivariate process by principal component analysis (PCA) of historical data. The technique uses a parallel coordinate system to visualize data that allows for monitoring of abnormal process events that lead to process faults, enabling the visualization of multiple principal components effectively and facilitating the study of how each principal component varies with respect to time. The principal component and residual space control limits are established for fault detection and the Random Forests machine learning tool is adopted for fault diagnosis. The key features of the methodology are demonstrated through a study of the benchmark Tennessee Eastman process.

A novel multi-dimensional visualization technique for understanding the design parameters of drug formulations

The quality-by-design concept is a new regulatory paradigm for pharmaceutical development, while the response surface method (RSM) is a promising approach for understanding design parameters for drug formulations. RSM aims to provide a visual image to support statistical design and analysis of experiments. However, neither contour plots nor 3D surface plots that have commonly been used can completely visualize interactions between the parameters within the design space, due to their limited dimensionality. This article presents a visualization technique that can simultaneously display the responses to multi-dimensional factors by mapping N-dimensional data onto unique x– y coordinates, re-defined by recursive slice-and-dice subdivision of the 2D plane. The applicability of the technique was confirmed using published data on the design of nasal drug formulations.

Multidimensional visualization and clustering for multiobjective optimization of artificial satellite heat pipe design

This study presents a newly developed approach for visualization of Pareto and quasi-Pareto solutions of a multiobjective design problem for the heat piping system in an artificial satellite. Given conflicting objective functions, multiobjective optimization requires both a search algorithm to find optimal solutions and a decision-making process for finalizing a design solution. This type of multiobjective optimization problem may easily induce equally optimized multple solutions such as Pareto solutions, quasi-Pareto solutions, and feasible solutions. Here, a multidimensional visualization and clustering technique is used for visualization of Pareto solutions. The proposed approach can support engineering decisions in the design of the heat piping system in artificial satellites. Design considerations for heat piping system need to simultaneously satisfy dual conditions such as thermal robustness and overall limitation of the total weight of the system. The proposed visualization and clustering technique can be a valuable design tool for the heat piping system, in which reliable decision-making has been frequently hindered by the conflicting nature of objective functions in conventional approaches.

Adjoint-based sensitivity analysis of fluid-structure interaction vibrations

Adjoint-based sensitivity analysis for the cavity fluid-structure interaction vibrations problem, with the inclusion of prestress and internal pressurization, is considered. The formulation employs a continuum approach in which the adjoint model is defined by employing a new duality relation between the sensitivities and the adjoint variables. For complex problems, such as fluid-structure interaction, the new duality relation provides an unambiguous approach for the construction of the adjoint variational problem in contrast to the standard approach. It also provides useful insight into the interpretation of the adjoint method. The resulting adjoint formulation for sensitivities is based on evaluation of inner products of the adjoint and state variables and is efficient because it requires the calculation of the adjoint field only once, regardless of the number of decision variables. In order to assess the accuracy of the adjoint approach for sensitivities in fluid-structure interaction problems, a model problem is considered and exact and analytical adjoint sensitivities are compared. The method is then applied to a finite element human eye model to compute the sensitivities with respect to elasticity constants and intraocular pressure for application to inverse estimation of those properties. The results are presented using a particularly effective multidimensional visualization technique.

Using table lens to interactively build classifiers

Rather than induce classification rules by sophisticated algorithms, we introduce a fully interactive approach for building classifiers from large multivariate datasets based on the table lens, a multidimensional visualization technique, and appropriate interaction capabilities. Constructing classifiers is an interaction with a feedback loop. The domain knowledge and human perception can be profitably included. In our approach, both continuous and categorical attributes are processed uniformly, and continuous attributes are partitioned on the fly. Our performance evaluation with data sets from the UCI repository demonstrates that this interactive approach is useful to easily build understandable classifiers with high prediction accuracy and no required a prior knowledge about the datasets.