Distance-based Similarity Measures Research Articles

A Few Similarity Measures on the Class of Trapezoidal-Valued Intuitionistic Fuzzy Numbers and Their Applications in Decision Analysis

Similarity measures on trapezoidal-valued intuitionistic fuzzy numbers (TrVIFNs) are functions that measure the closeness between two TrVIFNs, which has a lot of applications in the area of pattern recognition, clustering, decision-making, etc. Researchers around the world are proposing various similarity measures on the generalizations of fuzzy sets. However, many such measures do not satisfy the condition that “the similarity between two fuzzy numbers is equal to 1 implies that both the fuzzy numbers are equal” and this gives a pathway for the researchers to introduce different similarity measures on various classes of fuzzy sets. Also, all of them try to find out the similarity by using a single function, and in the present study, we try to propose a combined similarity measure principle by using four functions (four similarity measures). Thus, the main aim of this work is to introduce a few sets of similarity measures on the class of TrVIFNs and propose a combined similarity measure principle on TrVIFNs based on the proposed similarity measures. To do this, in this paper, firstly, we propose four distance-based similarity measures on TrVIFNs using score functions on TrVIFNs and study their mathematical properties by establishing various propositions, theorems, and illustrations, which is achieved by using numerical examples. Secondly, we propose the idea of a combined similarity measure principle by using the four proposed similarity measures sequentially, which is a first in the literature. Thirdly, we compare our combined similarity measure principle with a few important similarity measures introduced on various classes of fuzzy numbers, which shows the need for and efficacy of the proposed similarity measures over the existing methods. Fourthly, we discuss the trapezoidal-valued intuitionistic fuzzy TOPSIS (TrVIF-TOPSIS) method, which uses the proposed combined similarity measure principle for solving a multi-criteria decision-making (MCDM) problem. Then, we discuss the applicability of the proposed modified TrVIF-TOPSIS method by solving a model problem. Finally, we discuss the sensitivity analysis of the proposed approaches by using various cases.

DeepION: A Deep Learning-Based Low-Dimensional Representation Model of Ion Images for Mass Spectrometry Imaging.

Mass spectrometry imaging (MSI) is a high-throughput imaging technique capable of the qualitative and quantitative in situ detection of thousands of ions in biological samples. Ion image representation is a technique that produces a low-dimensional vector embedded with significant spectral and spatial information on an ion image, which further facilitates the distance-based similarity measurement for the identification of colocalized ions. However, given the low signal-to-noise ratios inherent in MSI data coupled with the scarcity of annotated data sets, achieving an effective ion image representation for each ion image remains a challenge. In this study, we propose DeepION, a novel deep learning-based method designed specifically for ion image representation, which is applied to the identification of colocalized ions and isotope ions. In DeepION, contrastive learning is introduced to ensure that the model can generate the ion image representation in a self-supervised manner without manual annotation. Since data augmentation is a crucial step in contrastive learning, a unique data augmentation strategy is designed by considering the characteristics of MSI data, such as the Poisson distribution of ion abundance and a random pattern of missing values, to generate plentiful ion image pairs for DeepION model training. Experimental results of rat brain tissue MSI show that DeepION outperforms other methods for both colocalized ion and isotope ion identification, demonstrating the effectiveness of ion image representation. The proposed model could serve as a crucial tool in the biomarker discovery and drug development of the MSI technique.

Utilizing the Clustering Techniques using Distance-Based Similarity Measures of SVNS in Medical Diagnosis

The clustering techniques, combined with distance-based similarity measures of Single Valued Neutrosophic Sets (SVNS) are studied and applied in medical diagnosis. The study starts with reviewing SVNS' theoretical foundations, emphasising its ability to capture and handle ambiguous data. This study focuses on integrating distance-based similarity measurements to improve the clustering process, which has seen limited implementation thus far. The set of data includes three patients with five symptoms and three diagnoses. To deal with the data in medical diagnosis, each patient is diagnosed with a disease based on distance-based similarity measures. The disease with the highest similarity measure value indicates the recognized disease for that patient. Then, the diseases are clustered into different categories depend on the values of confidence level. The obtained results show that the suggested method enhances the precision of medical diagnosis significantly, especially in cases with ambiguity and uncertainty.

Improved Whale Optimization Algorithm with Deep Learning-Driven Retinal Fundus Image Grading and Retrieval

Several Deep Learning (DL) and medical image Machine Learning (ML) methods have been investigated for efficient data representations of medical images, such as image classification, Content-Based Image Retrieval (CBIR), and image segmentation. CBIR helps medical professionals make decisions by retrieving similar cases and images from electronic medical image databases. CBIR needs expressive data representations for similar image identification and knowledge discovery in massive medical image databases explored by distinct algorithmic methods. In this study, an Improved Whale Optimization Algorithm with Deep Learning-Driven Retinal Fundus Image Grading and Retrieval (IWOADL-RFIGR) approach was developed. The presented IWOADL-RFIGR method mainly focused on retrieving and classifying retinal fundus images. The proposed IWOADL-RFIGR method used the Bilateral Filtering (BF) method to preprocess the retinal images, a lightweight Convolutional Neural Network (CNN) based on scratch learning with Euclidean distance-based similarity measurement for image retrieval, and the Least Square Support Vector Machine (LS-SVM) model for image classification. Finally, the IWOA was used as a hyperparameter optimization technique to improve overall performance. The experimental validation of the IWOADL-RFIGR model on a benchmark dataset exhibited better performance than other models.

A novel image recognition using Fuzzy C-Means and content-based fabric image retrieval

ABSTRACT Due to the growth of image exploitation, there is a requirement of systemizing the images in several ways. Time consumption and accuracy are the most important problem in the CBIR method. Therefore, a Content-Based Fabric Image Retrieval (CBFIR) method is proposed in this paper that is based on texture and colour features extraction. The Fuzzy C-Means (FCM) algorithm creates primary clusters and the Simultaneous Orthogonal Matching Pursuit (SOMP) algorithm updates every cluster in the dictionary. To find the sparse representation dictionary, the test image is compared by the proposed method with the dictionaries generated. The input images are obtained from the fabric dataset. Several distance based similarity measures are utilized for comparison. It can be seen that the proposed method’s performance is sufficiently good concerning the rate of recall, accuracy, and precision with the values of 97%, 91%, and 88% respectively.

Multimodal biomedical image retrieval and indexing system using handcrafted with deep convolution neural network features

The advances in biomedical imaging equipment have produced a massive amount of medical images that are generated by the different modalities. Consequently, a huge volume of data has been produced and caused a complex and time-consuming retrieving process of the relevant cases. To resolve this issue, the Content-Based Biomedical Image Retrieval (CBMIR) system is applied to retrieve the related images from the earlier patients’ databases. However, the previous handcrafted features methods that applied the CBMIR model have shown poor performance in many multimodal databases. In this paper, we focus on designing CBMIR technique using Deep Learning (DL) models. We present a new Multimodal Biomedical Image Retrieval and Classification (M-BMIRC) technique for retrieving and classifying the biomedical images from huge databases. The proposed M-BMIRC model involves three dissimilar processes as following: feature extraction, similarity measurement, and classification. It uses an ensemble of handcrafted features from Zernike Moments (ZM) and deep features from Deep Convolutional Neural Networks (DCNN) for feature extraction process. Additionally, the Hausdorff Distance based similarity measure is employed to identify the resemblance between the queried image and the images that exist in the database. Moreover, the classification process gets executed on the retrieval images using the Probabilistic Neural Network (PNN) model, which allocates the class labels of the tested images. Finally, the experimental studies are conducted using two benchmark medical datasets and the results ensure the superior performance of the proposed model in terms of different measures include Average Precision Rate (APR), Average Recall Rate (ARR), F-score, accuracy, and Computation Time (CT).

Fuzzy fault tree analysis of chlorine gas release hazard in Chlor-Alkali industry using [formula omitted]-cut interval-based similarity aggregation method

In chemical industries, the release of toxic gases and chemicals is always a serious concern as it has the potential to harm human life and the environment, leading to significant losses. Risk assessment in chemical industries is essential to identify and prevent such type of unwanted hazards. Conventional fault tree analysis has been used effectively to identify the system failure causes and evaluate the system reliability, but it requires quantitative historical failure data of system components. In many real-world applications, it can be difficult to obtain precise and sufficient quantitative failure data of system components. This research paper proposes a novel fuzzy set theory-based fault tree analysis to evaluate system failure probability by utilizing the available qualitative data such as expert opinions/views when quantitative historical failure data of system components is unavailable or insufficient. To aggregate the different opinions of selected experts, a new similarity aggregation method is developed by using interval distance-based similarity measure function. The proposed fuzzy fault tree analysis applies system fault tree, α-cut intervals of fuzzy membership functions, and interval arithmetic operations. To validate the applicability and effectiveness of proposed approach, the basic event probabilities in fault tree analysis of chlorine release from storage and filling facility are generated using proposed approach and then compared with the known probabilities. Further, importance analysis is performed to evaluate the contribution of each basic event in top event “chlorine release” occurrence that would help the decision makers to identify the risk factor of chlorine release. The obtained results may be helpful for decision makers to decide whether and where to take protective actions in the risk management process.

Fault diagnosis for lithium-ion batteries in electric vehicles based on signal decomposition and two-dimensional feature clustering

Battery fault diagnosis is essential for ensuring the reliability and safety of electric vehicles (EVs). The existing battery fault diagnosis methods are difficult to detect faults at an early stage based on the real-world vehicle data since lithium-ion battery systems are usually accompanied by inconsistencies, which are difficult to distinguish from faults. A fault diagnosis method based on signal decomposition and two-dimensional feature clustering is introduced in this paper. Symplectic geometry mode decomposition (SGMD) is introduced to obtain the components characterizing battery states, and distance-based similarity measures with the normalized extended average voltage and dynamic time warping distances are established to evaluate the state of batteries. The 2-dimensional feature clustering based on DBSCAN is developed to reduce the number of feature thresholds and differentiate flaw cells from the battery pack with only one parameter under a wide range of values. The proposed method can achieve fault diagnosis and voltage anomaly identification as early as 43 days ahead of the thermal runaway. And the results of four electric vehicles and the comparison with other traditional methods validated the proposed method with strong robustness, high reliability, and long time scale warning, and the method is easy to implement online.

Development of a hat style recognition system based on image processing and machine learning

The purpose of this paper is to investigate the recognition mechanism of hat styles and develop a corresponding hat style recognition system (HSRS). An image processing and machine learning integrated method (IPML) is proposed and validated for automatic hat style recognition. First, 4 kinds of hat styles (borderless knitted hats, berets, top hats and peaked hats) with 800 pictures are employed as research objects and divided into two categories: the first 400 serve as the training set and the rest 400 as the test set. Then, IPML is proposed to obtain a hat silhouette. Curvature feature points are extracted from hat silhouette and further used as parameters for the automatic hat style recognition. In the real recognition process, a new case is compared with the pre-set 400 samples in the training set regarding these characteristic parameters. A Hausforff distance-based similarity measurement tool is used in the comparison process. The experimental results show that when the curvature feature points are 70 and the output results are 3, the average recognition accuracy rate can reach 90.5%, of which the value of borderless knitted hats is the highest with 98% and followed by the top hats with 95%. This work can be used for hat recommendation systems. It can also be extended to support the area of personalized industrial product design such as fashion design, furniture design and advertisement design.

A Novel Similarity Measure of Single-Valued Neutrosophic Sets Based on Modified Manhattan Distance and Its Applications

A single-valued neutrosophic (SVN) set contains three parameters, which can well describe three aspects of an objective thing. However, most previous similarity measures of SVN sets often encounter some counter-intuitive examples. Manhattan distance is a well-known distance, which has been applied in pattern recognition, image analysis, ad-hoc wireless sensor networks, etc. In order to develop suitable distance measures, a new distance measure of SVN sets based on modified Manhattan distance is constructed, and a new distance-based similarity measure also is put forward. Then some applications of the proposed similarity measure are introduced. First, we introduce a pattern recognition algorithm. Then a multi-attribute decision-making method is proposed, in which a weighting method is developed by building an optimal model based on the proposed similarity measure. Furthermore, a clustering algorithm is also put forward. Some examples are also used to illustrate these methods.

Density clustering with divergence distance and automatic center selection

The density peak clustering (DPC) algorithm is a famous density-based method for exploring, analyzing, and deriving information from data. In the case of application, it encounters the following bottlenecks. Firstly, its Euclidean distance based similarity measurement is prone to misclassification of neighbors. Secondly, its clustering results are significantly influenced by human factors (controlling the cut-off distance parameter dc and selecting the clustering center manually). Finally, the local density ρ is affected by parameter dc and cannot reflect the sparseness of the data distribution. To overcome these deficiencies, a novel density clustering algorithm, called NAPC, is proposed. First, a Divergence distance is defined to evaluate the similarity of points under the refined Euclidean distance space. Second, the specific value of dc is obtained based on the theory of Adjusted Boxplot. Then, the local density ρ is reinterpreted based on the above Divergence distance and newly assigned dc. Finally, a judgement index is given to determine which points can be regarded as the center points in the Divergence distance space. The performance of our proposal is evaluated by comparing with several existing clustering methods on various datasets. Experimental results prove that NAPC performs much better than several comparison algorithms. It can identify clusters of various shapes and spatial dimensions with minimal human intervention.

An Enhanced Re-Ranking Model for Person Re-Identification

Presently, Person Re-IDentification (PRe-ID) acts as a vital part of real time video surveillance to ensure the rising need for public safety. Resolving the PRe-ID problem includes the process of matching observations of persons among distinct camera views. Earlier models consider PRe-ID as a unique object retrieval issue and determine the retrieval results mainly based on the unidirectional matching among the probe and gallery images. But the accurate matching might not be present in the top-k ranking results owing to the appearance modifications caused by the difference in illumination, pose, viewpoint, and occlusion. For addressing these issues, a new Hyper-parameter Optimized Deep Learning (DL) approach with Expanded Neighborhood Distance Reranking (HPO-DLDN) model is proposed for PRe-ID. The proposed HPO-DLDN involves different processes for PRe-ID, such as feature extraction, similarity measurement, and feature re-ranking. The HPO-DLDN model uses a Adam optimizer with Densely Connected Convolutional Networks (DenseNet169) model as a feature extractor. Additionally, Euclidean distance-based similarity measurement is employed to determine the resemblance between the probe and gallery images. Finally, the HPO-DLDN model incorporated ENDR model to re-rank the outcome of the person-reidentification along with Mahalanobis distance. An extensive experimental analysis is carried out on CUHK01 benchmark dataset and the obtained results verified the effective performance of the HPO-DLDN model in different aspects.

Optimization of Fuzzy Support Vector Machine (FSVM) Performance by Distance-Based Similarity Measure Classification

This research aims to determine the maximum or minimum value of a Fuzzy Support Vector Machine (FSVM) Algorithm using the optimization function. SVM is considered as an effective method of data classification, as opposed to FSVM, which is less effective on large and complex data because of its sensitivity to outliers and noise. One of the techniques used to overcome this inefficiency is fuzzy logic with its ability to select the right membership function, which significantly affects the effectiveness of the FSVM algorithm performance. This research was carried out using the Gaussian membership function and the Distance-Based Similarity Measurement consisting of the Euclidean, Manhattan, Chebyshev, and Minkowsky distance methods. Subsequently, the optimization of the FSVM classification process was determined using four proposed FSVM models and normal SVM as comparison references. The results showed that the method tends to eliminate the impact of noise and enhance classification accuracy effectively. FSVM provides the best and highest accuracy value of 94% at a penalty parameter value of 1000 using the Chebyshev distance matrix. Furthermore, the model proposed will be compared to the performance evaluation model in preliminary studies (Xiao Kang et al., 2018). The result further showed that using FSVM with Chebyshev distance matrix and a Gaussian membership function provides a better performance evaluation value. Doi: 10.28991/HIJ-2021-02-04-02 Full Text: PDF

Vision-based vehicle tracking on highway traffic using bounding-box features to extract statistical information

In this study, a new bounding-box based vehicle tracking algorithm is presented to extract statistical information in the highway traffic. A novel shaking filter and a new voting approach are employed in the vehicle detection and tracking phases to reduce camera shaking effects that cause misdetection, misclassification, and mistracking. The algorithm uses image streams captured via ordinary cameras and successfully classifies and determines the time-dependent vehicle trajectory through successive frames. The novel tracking algorithm utilizes the Euclidean distance-based similarity measure to associate the detected vehicles in successive frames, or it predicts the next state of vehicles using the linear/polynomial prediction functions obtained from the trajectory vector when the observed vehicles are disappeared from the scene due to the occlusion or illusion problems. The comparative vehicle counting results show that the proposed algorithm performs approximately 7% better than the Kalman filter-based tracker.

Robot Selection Problem via Fuzzy TOPSIS Method Using Novel Distance and Similarity Measure for Generalized Fuzzy Numbers with Unequal Heights

Background: Mega multinational companies are highly dependent on robots to handle the maximum of their machinery workload, which significantly reduces human labor and saves valuable time as well. However, as vital as the role of robots is, a much more challenging task is its selection. Moreover, the robots need to be evaluated on the grounds of different specifications and their ease of handling, which results in a smooth and work-efficient environment. Objective: The prime objective of this paper is to devise a fruitful decision-making model for a robot selection problem, which utilizes a multi-criteria decision-making method known as Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The TOPSIS method is based on the newly defined distance measure involving generalized fuzzy numbers with unequal heights (GFNUHs). Methodology/Approach: At first, we define a novel distance measure based on the “expected value” and “variance” of GFNUHs, where both the parameters are evaluated with the help of the [Formula: see text]-cut method. We then also give the expression for the distance-based similarity measure and investigate some of their properties. Both the distance and the similarity measure(s) are then validated for their effectiveness through a hypothetical case study of pattern recognition. Moreover, we consider 10 different bunches of generalized fuzzy numbers (GFNs) and present a comparative study with the already established measures to establish the efficiency and superiority of our proposed measures. Finally, the distance measure is deployed in the TOPSIS method, which facilitates suitable robot selection by an automobile company. Findings/Results: A comparison of results for the proposed distance measure and the similarity measure with the existing ones is presented which proves that the proposed measure(s) are effective and usable. Novelty/Value: The evaluation of expected value and variance of GFNUHs with the help of [Formula: see text]-cut technique is a completely original idea showcased in this paper and its improved version of TOPSIS for GFNUHs as discussed shall add a new direction in the realm of decision-making.

Medical Diagnosis via Distance-based Similarity Measure for Rough Neutrosophic Set

Medical Diagnosis via Distance-based Similarity Measure for Rough Neutrosophic Set

Time Series Clustering with Topological and Geometric Mixed Distance

Time series clustering is an essential ingredient of unsupervised learning techniques. It provides an understanding of the intrinsic properties of data upon exploiting similarity measures. Traditional similarity-based methods usually consider local geometric properties of raw time series or the global topological properties of time series in the phase space. In order to overcome their limitations, we put forward a time series clustering framework, referred to as time series clustering with Topological-Geometric Mixed Distance (TGMD), which jointly considers local geometric features and global topological characteristics of time series data. More specifically, persistent homology is employed to extract topological features of time series and to compute topological similarities among persistence diagrams. The geometric properties of raw time series are captured by using shape-based similarity measures such as Euclidean distance and dynamic time warping. The effectiveness of the proposed TGMD method is assessed by extensive experiments on synthetic noisy biological and real time series data. The results reveal that the proposed mixed distance-based similarity measure can lead to promising results and that it performs better than standard time series analysis techniques that consider only topological or geometrical similarity.

Euclidean Distance Based Similarity Measurement and Ensuing Ranking Scheme for Document Search from Outsourced Cloud Data

In this paper, we propose the Euclidean Distance based Similarity Measurement and Ensuing Ranking (EDSMER) scheme to aid effective document search from outsourced cloud data. It is another attempt to find an alternative to binary based approaches. In this approach, the User or the Data owner needs to filter out the suitable keywords for the document and then the index is prepared. To provide security and privacy, both the data and the index are encrypted and moved to the cloud space. The application of Euclidean Distance based Similarity Measurement and Ensuing Ranking (EDSMER) scheme for document searching takes place after the authorized user requests for the documents through query terms. Initially the authorized user sends a query to Cloud Service Provider to retrieve all the documents which are mapped with the keywords provided by him. The proposed algorithm calculates the distance between the query terms and the index terms. The minimum the distance, the more it is closer towards each other and vice-versa. Our Euclidean Distance based Similarity Measurement and Ensuing Ranking (EDSMER) scheme greatly enhances the system functionality by sending the most relevant documents instead of transmitting all documents back. The experimental validations are performed on RFC and FIRE dataset. Through experimental analysis, we prove that our proposed approach is secure and efficient as well as exhibits better recall and precision rate in the IR system to deal with the document-retrieval process.

Ensembling of text and images using Deep Convolutional Neural Networks for Intelligent Information Retrieval

Information retrieval (IR) defines the process of searching and attaining specific information resources which are related to the specific information requirements from the available resource pool. It finds useful in several real time application areas namely digital library, healthcare, education, internet browsing, etc. Recently, deep learning (DL) models become popular in different fields of image processing, object detection, and natural language processing. Therefore, in this paper, DL models are employed to retrieve the text and images proficiently. This paper presents an ensemble of DL based IR models for text and images. The proposed model intends to develop DL models individually for text and images. Initially, convolutional neural network based VGGNet-19 model is used as a feature extractor and Euclidian distance based similarity measurement for the retrieval of images. At the same time, bidirectional-long short-term memory (BiLSTM) technique is applied for retrieval of textual documents. The presented BiLSTM model sequentially considers every word in a sentence, extracts the details and embeds it to the semantic vector. In addition to the feature extraction using deep learning techniques, the similarity measurement emphasis the closeness of the document to the given query. The proposed retrieval system has tested on text and images for both general and specific domain (agriculture) with the datasets of Yahoo, Google and Corel10K. With the datasets the performance has been computed by the standard measures such as precision, recall and F-score where the proposed deep learning model produces better results when compared to existing techniques. The proposed model has been tested for the specific domain and achieves the performance of 93% precision and 85% recall and 90% F-score when compared to the existing model.

Geospatial Analytics for COVID-9 Active Case Detection

Ever since the COVID-19 pandemic started in Wuhan, China, much research work has been focusing on the clinical aspect of SARS-CoV-2. Researchers have been leveraging on various Artificial Intelligence techniques as an alternative to medical approach in understanding the virus. Limited studies have, however, reported on COVID-19 transmission pattern analysis, and using geography features for prediction of potential outbreak sites. Predicting the next most probable outbreak site is crucial, particularly for optimizing the planning of medical personnel and supply resources. To tackle the challenge, this work proposed distance-based similarity measures to predict the next most probable outbreak site together with its magnitude, when would the outbreak likely to happen and the duration of the outbreak. The work began with preprocessing of 1365 patient records from six districts in the most populated state named Selangor in Malaysia. The dataset was then aggregated with population density information and human elicited geography features that might promote the transmission of COVID-19. Empirical findings indicated that the proposed unified decision-making approach outperformed individual distance metric in predicting the total cases, next outbreak location, and the time interval between start dates of two similar sites. Such findings provided valuable insights for policymakers to perform Active Case Detection.