Clustering-based Feature Selection Research Articles

Feature Selection Based on Unsupervised Clustering Mechanism on Multiple-Sequence MRIs for Predicting Neoadjuvant Chemoradiation Response in Locally Advanced Rectal Cancer

Accurate response prediction allows for personalized cancer management. We developed an unsupervised clustering mechanism to improve effectiveness and efficiency in feature selection operation for accurate patient stratification. Forty-three locally advanced rectal cancer (LARC) patients underwent neoadjuvant chemoradiation were included, pre-treatment T2 and ADC MRIs were acquired for each patient. An initial feature space consisting of 200 radiomic features extracted from manually delineated GTVs from two sequences of MR images. Additional 960 high-order radiomic features extracted from a 3D convolutional neural network (CNN). To remove redundant and irrelevant features, we developed an unsupervised clustering-based feature selection operation to determine the combination of features with potential best performance. The normal process of feature selection involves searching new feature combinations and training new classifiers for evaluating their performance via an iterative process based on selected feature set, the overall time cost is tremendous. To balance the computational cost and search efficiency, firstly, we proposed an unsupervised clustering analysis metric- Comprehensive Cluster Analysis Index (CCAI) through the K-means algorithm, where the average distances between the sample points and the cluster centroids and so on, to construct a multiple linear regression model. Secondly, we extracted sample points by varying the number of features and feature ratios between radiomic features and 3D-CNN features in the output of feature selection. Thirdly, we optimized the model using the sampling points to calculate the CCAI. Two typical feature combination search algorithms, the random forest recursive feature elimination (RF-RFE) and the differential evolution (DE), were used to perform feature selection with CCAI. The accuracy, area-under-curve (AUC) and specificity, based on combined 3D-CNN and radiomic features extracted from combined T2 and ADC images, were 0.852, 0.871, and 0.735, respectively. Our experiments illustrated higher predictive power (AUC = 0.846) based on high-order abstract features extracted from the CNN on ADC and T2 images, compared to the traditional radiomic model (AUC = 0.714). Additionally, the predictive models constructed based on radiomics and CNN features extracted from ADC images were more predictable in terms of treatment responses than the radiomic and CNN imaging features extracted from T2 images. The average computational time of DE and RF-RFE were 50.5s and 128.6s in one single computation, the average computational time were 24.2s and 91.3s with CCAI, respectively. We proposed an unsupervised clustering analysis mechanism to improve the effectiveness of feature selection while decreasing its time cost markedly, which highlight the correlation and complementarity between low- and high-level imaging features, achieving better predictive accuracy.

Identifying cancer risks using spectral subset feature selection based on multi-layer perception neural network for premature treatment

Recently, human beings have been affected mainly by dreadful cancer diseases. Predicting cancer risk levels is a major challenge in biomedical research for feature selection and classification at the margins. To resolve this problem, we propose a Subset Clustering-Based Feature Selection using a Multi-Layer Perception Neural Network (SCFS-MLPNN). Initially, pre-processing is carried out with Intensive Mutual Disease Influence Rate (IMDIR) to identify the relational features. In addition, the Successive Disease Pattern Stimulus Rate (SDPSR) is carried out to create relative feature patterns. Based on the patterns, the features are selected and grouped into clustering. Inter-Class Sub-Space Clustering (ICSSC) is applied to split the features by class labels depending on the marginal rate. From the class labels, marginal features are obtained using spectral subset feature selection (SSFS). The selected features are then trained in a Multi-Layer Perception Neural Network (MLPNN) classifier to classify the patient features by risk. Its contribution is to exploit subset features to improve classification accuracy by clustering relational features. The proposed classifier yields higher classification accuracy than previous methods and observes cancer detection for early detection. Therefore, the proposed method achieved a risk analysis accuracy of 91.8% and an F-measure of 91.3% for early detection, which is recommended for early diagnosis.

A Clustering based Feature Selection Approach using Maximum Spanning Tree

Mutual information (MI) based feature selection methods are getting popular as its ability to capture the nonlinear and linear relationship among random variables and thus it performs better in different fields of machine learning. Traditional MI based feature selection algorithms use different techniques to find out the joint performance of features and select the relevant features among them. However, to do this, in many cases, they might incorporate redundant features. To solve these issues, we propose a feature selection method, namely Clustering based Feature Selection (CbFS), to cluster the features in such a way so that redundant and complementary features are grouped in the same cluster. Then, a subset of representative features is selected from each cluster. Experimental results of CbFS and four state-of-the-art methods are reported to measure the excellency of CbFS over twenty benchmark UCI datasets and three renowned network intrusion datasets. It shows that CbFS performs better than the comparative methods in terms of accuracy and performs better in identifying attack or normal instances in security datasets. DUJASE Vol. 7 (2) 47-55, 2022 (July)

UFODMV: Unsupervised Feature Selection for Online Dynamic Multi-Views

In most machine learning (ML) applications, data that arrive from heterogeneous views (i.e., multiple heterogeneous sources of data) are more likely to provide complementary information than does a single view. Hence, these are known as multi-view data. In real-world applications, such as web clustering, data arrive from diverse groups (i.e., sets of features) and therefore have heterogeneous properties. Each feature group is referred to as a particular view. Although multi-view learning provides complementary information for machine learning algorithms, it results in high dimensionality. However, to reduce the dimensionality, feature selection is an efficient method that can be used to select only the representative features of the views so to reduce the dimensionality. In this paper, an unsupervised feature selection for online dynamic multi-views (UFODMV) is developed, which is a novel and efficient mechanism for the dynamic selection of features from multi-views in an unsupervised stream. UFODMV consists of a clustering-based feature selection mechanism enabling the dynamic selection of representative features and a merging process whereby both features and views are received incrementally in a streamed fashion over time. The experimental evaluation demonstrates that the UFODMV model has the best classification accuracy with values of 20% and 50% compared with well-known single-view and multi-view unsupervised feature selection methods, namely OMVFS, USSSF, and SPEC.

Optimal cluster based feature selection for intrusion detection system in web and cloud computing environment using hybrid teacher learning optimization enables deep recurrent neural network

Information technology organizations have experienced rapid growth in recent years, resulting in scalability, mobility, and flexibility challenges. Those organizations move their data to the cloud because security and privacy are major concerns. As cloud computing becomes more popular, security has become an important concern. These confidential data are vulnerable to attacks/malicious or intruders due to the characteristics of the cloud. In order to address the growing concern of real-time intruders, a variety of intrusion detection systems (IDS) used specifically for cloud environments with the aim of enhancing overall security. There are, however, some limitations and known attacks that can be overcome by those IDSs. We recently proposed a hybrid soft computing based IDS (ST-IDS) for web and cloud environments, but missed some novel web and cloud attacks. Using hybrid teacher learning enabled deep recurrent neural networks and cluster based feature optimization, we propose an IDS scheme for web and cloud computing environments. MMFO (modified manta-ray foraging optimization) is used after feature extraction to select optimal features for further detection. To classify the intrusion in the web-cloud environment, a hybrid teacher-learning enabled deep recurrent neural network (TL-DRNN) is introduced. Our proposed IDS scheme has been validated using benchmark datasets including DARPA LLS DDoS-1.0, CICIDS-2017, and CSIC-2010. The performance of our proposed IDS scheme has been compared to existing IDS schemes using various quality measures.

Synergism of Multi-Modal Data for Mapping Tree Species Distribution—A Case Study from a Mountainous Forest in Southwest China

Accurately mapping tree species is crucial for forest management and conservation. Most previous studies relied on features derived from optical imagery, and digital elevation data and the potential of synthetic aperture radar (SAR) imagery and other environmental factors have, generally, been underexplored. Therefore, the aim of this study is to evaluate the potential of fusing freely available multi-modal data for accurately mapping tree species. Sentinel-2, Sentinel-1, and various environmental datasets over a large mountainous forest in Southwest China were obtained and analyzed using Google Earth Engine (GEE). Seven data cases considering the individual or joint performance of different features, and four additional cases considering a novel clustering-based feature selection method, were analyzed. All 11 cases were assessed using three machine learning algorithms, including random forest (RF), support vector machine (SVM), and extreme gradient boosting tree (XGBoost). The best performance, with an overall accuracy of 77.98%, was attained from the case with all features and the random forest classifier. Sentinel-2 data alone exhibited similar performance as environmental data in terms of overall accuracy. Similar species, such as oak and birch, cannot be spectrally discriminated based on Sentinel-2-based features alone. The addition of SAR features improved discrimination, especially when distinguishing between some coniferous and deciduous species, but also decreased accuracy for oak. The analysis based on different data cases and feature importance rankings indicated that environmental features are important. The random forest outperformed other models, and a better prediction was achieved for planted tree species compared to that for the natural forest. These results suggest that accurately mapping tree species over large mountainous areas is feasible with freely accessible multi-modal data, especially when considering environmental factors.

A Metaheuristic Technique for Cluster-Based Feature Selection of DNA Methylation Data for Cancer

Epigenetics is the study of phenotypic variations that do not alter DNA sequences. Cancer epigenetics has grown rapidly over the past few years as epigenetic alterations exist in all human cancers. One of these alterations is DNA methylation; an epigenetic process that regulates gene expression and often occurs at tumor suppressor gene loci in cancer. Therefore, studying this methylation process may shed light on different gene functions that cannot otherwise be interpreted using the changes that occur in DNA sequences. Currently, microarray technologies; such as Illumina Infinium BeadChip assays; are used to study DNA methylation at an extremely large number of varying loci. At each DNA methylation site, a beta value (β) is used to reflect the methylation intensity. Therefore, clustering this data from various types of cancers may lead to the discovery of large partitions that can help objectively classify different types of cancers as well as identify the relevant loci without user bias. This study proposed a Nested Big Data Clustering Genetic Algorithm (NBDC-GA); a novel evolutionary metaheuristic technique that can perform cluster-based feature selection based on the DNA methylation sites. The efficacy of the NBDC-GA was tested using real-world data sets retrieved from The Cancer Genome Atlas (TCGA); a cancer genomics program created by the National Cancer Institute (NCI) and the National Human Genome Research Institute. The performance of the NBDC-GA was then compared with that of a recently developed metaheuristic Immuno-Genetic Algorithm (IGA) that was tested using the same data sets. The NBDC-GA outperformed the IGA in terms of convergence performance. Furthermore, the NBDC-GA produced a more robust clustering configuration while simultaneously decreasing the dimensionality of features to a maximum of 67% and of 94.5% for individual cancer type and collective cancer, respectively. The proposed NBDC-GA was also able to identify two chromosomes with highly contrasting DNA methylations activities that were previously linked to cancer.

Semi-supervised Clustering Based Feature Selection with Multiobjective Genomic Search Class-based Classification Method for NIDPS

Objectives: The purpose of semi-supervised clustering-based feature selection with the multiobjective genomic search class based classification process is to extract the intrusion features from the content of the unbalanced class field and structures in the dataset. Method: A class-based taxonomy with Multiobjective Genomic Search Method (SCMGSM) is semi-supervised Clustering Based Feature Selection designed to avoid inappropriate classification and to generate class based efficient authenticity for detecting intrusions. The proposed SCMGSM operates on a cluster class basis to reduce the overhead distribution by utilizing the subset classifications based on labelling process. For cluster class based feature selection there is not necessary to retain the interconnected features, so the proposed taxonomy to detect the attack features for each subset classification improves the performance. SCMGSM protects the data source using the potential blocking and minimizes the hassle of unauthorized users from changing features. Findings: The experiments focused on a set of data to identify and classify features based on each class. The proposed SCMGSM handles enhanced classification accuracy with 5.37% normal class completeness, reduces false positive rate to 2.78%, improves detection rate to 6.75% and reduces the backup issue by 10.65% compared to overhead classification. Novelty: SCMGSM operates based on multi objective genomic search based cluster class taxonomy and offers high degree of accuracy, classification and low false-positive rating compared to the detection of exploitation, Fuzzers, Generic and Renaissances attack classes with in a specific subset or a group of features. Keywords: Classification; Cluster; Feature selection; Intrusion detection; Network attacks

An intelligent feature selection approach with systolic tree structures for efficient association rules in big data environment

Frequent Pattern Mining (FPM) is regularly used in data mining applications for identifying objects of interest that frequent transactional databases. Traditional ARM algorithms work on minimum levels of support and confidence metrics which have to be defined and are subjective. This work attempts to solve this issue by proposing techniques for the automatic determination of these metrics. The Kernel Possibilistic Fuzzy Local Information C-Means (KPFLICM) Clustering and K-means Clustering-based Feature Selection algorithm for Association Rules (KCFSAR) is used where FPMs are based on Systolic Tree Structures (STS). These structures aim to achieve better accuracy by mimicking internal memory structures of the Frequent Pattern Growths (FPG) algorithm. Moreover, Chaotic Butterfly Optimization Method (CBOAs) is used to obtain the optimal support value and CBOA subsequently discovers global optimal solutions. This work's suggested scheme demonstrates its efficacy and superiority in experimental findings when compared to other approaches in terms of feature subset sizes, accuracies, execution times, and memory consumptions.

Reactive search-MST optimized clustering-based feature selection

Data clustering is a technique for analyzing the data that is incurred in various fields such as data processing, pattern recognition, knowledge discovery and machine learning. Feature clustering is an important paradigm for different types of feature selection techniques that aims to reduce redundant and irrelevant features from a given set of features in order to maintain load balance on the classification algorithm. The work proposed a PSO–GSO–MST, a hybrid approach that combines Particle Swarm Optimization (PSO) and Glowworm Swarm Optimization (GSO). The work performs efficient feature selection with improved classification accuracy. Clustering analysis plays an important role in knowledge discovery and data mining. It adopts the unsupervised learning method, and the results of clustering are similar within the class and are different between the classes. Aiming at some shortcomings of traditional clustering algorithms, some techniques for clustering using natural heuristic algorithms have emerged. The proposed work performs cluster using optimized Minimum Spanning Tree (MST). The work aims to perform optimization of MST with the help of two renowned techniques such as PSO and GSO. The proposed PSO–GSO–MST is compared with state-of-the-art algorithms such as Clustering-based Feature Selection (CFS) and PSO–MST. The results show that the classification accuracy for the proposed PSO–GSO–MST performs better by 16.9% than CFS and by 4.7% than PSO–MST optimized CFS, respectively. The outcome of the work proves that the proposed algorithm achieves improved performance than the currently available algorithms and can be used for clustering applications.

Texture classification using Minkowski distance measure-based clustering for feature selection

Texture classification has been one of the problem-solving strategies. The feature selection technique is critical for increasing the efficiency of machine learning because it significantly improves the performance of texture classification by removing irrelevant features from the original collection. The majority of methods for selecting features are statistical in nature. Accuracy in database classification is primarily achieved by selecting attributes and simultaneously increasing the classification rate. The task’s primary objective is to select the most significant features from the feature database for execution. We use the cluster-based feature selection method to derive features for statistical and structural approaches. The proposed strategy should be implemented in three stages. To begin, the function can be extracted using the local binary pattern (LBP), the gray level co-occurrence matrix, or the Gabor filter. Following that, a modified K-means clustering method is used to select features using four standard distance measurements: Euclidean, Minkowski, Chebychev, and City Block. Finally, benchmark classifiers such as naive bayes (NB), support vector machine (SVM), and K-nearest neighbor (KNN) were used to effectively compute the classification accuracy of the research. The proposed approach achieves accuracy of 99.4328 for KNN, 98.6654 for NB, and 99.6712 for SVM with the highest classification rate.

RFC: A feature selection algorithm for software defect prediction

Software defect prediction (SDP) is used to perform the statistical analysis of historical defect data to find out the distribution rule of historical defects, so as to effectively predictdefects in the new software. However, there are redundant and irrelevant features in the software defect datasets affecting the performance of defect predictors. In order to identify and remove the redundant and irrelevant features in software defectdatasets, we propose Relief F-based clustering (RFC), a cluster-based feature selection algorithm. Then, the correlation between features is calculated based on the symmetric uncertainty. According to the correlation degree, RFC partitions features into kclusters based on the k-medoids algorithm, and finally selects the representative features from each cluster to form the final feature subset. In the experiments, we compare the proposed RFC with classical feature selection algorithms on nine National Aeronautics and Space Administration (NASA) software defectprediction datasets in terms of area under curve (AUC) and F-value. The experimental results show that RFC can effectively improve the performance of SDP.

Clustering-based feature subset selection with analysis on the redundancy–complementarity dimension

In the era of big data, dimensionality reduction plays an extremely important role in many fields driven by machine learning and data mining techniques. The existing information-theoretic feature selection algorithms generally reduce the dimension by selecting the features with maximum class-relevance and minimum redundancy, while relatively overlook the complementary correlation among features and sometimes deal with it improperly. This paper proposes a novel feature subset selection algorithm called the Clustering-based Feature Selection with Redundancy–Complementarity Analysis (CFSRCA). The proposed algorithm can be mainly divided into two steps, namely, (a) selecting the candidate class-relevant features, and (b) selecting the representative features. In the latter step, the representative features are defined as the features with minimum redundancy and maximum complementarity, and a clustering method based on the minimum spanning tree (MST) is proposed to distinguish them effectively. To validate the effectiveness of CFSRCA, three comparative feature selection algorithms (ReliefF, CFS, and FOU) and four well-known classifiers (C4.5, SVM, kNN, and NBC) are used to conduct classification experiments on eight datasets. Experimental results verify the effectiveness of the proposed feature subset algorithm.

Multi-scale supervised clustering-based feature selection for tumor classification and identification of biomarkers and targets on genomic data

BackgroundThe small number of samples and the curse of dimensionality hamper the better application of deep learning techniques for disease classification. Additionally, the performance of clustering-based feature selection algorithms is still far from being satisfactory due to their limitation in using unsupervised learning methods. To enhance interpretability and overcome this problem, we developed a novel feature selection algorithm. In the meantime, complex genomic data brought great challenges for the identification of biomarkers and therapeutic targets. The current some feature selection methods have the problem of low sensitivity and specificity in this field.ResultsIn this article, we designed a multi-scale clustering-based feature selection algorithm named MCBFS which simultaneously performs feature selection and model learning for genomic data analysis. The experimental results demonstrated that MCBFS is robust and effective by comparing it with seven benchmark and six state-of-the-art supervised methods on eight data sets. The visualization results and the statistical test showed that MCBFS can capture the informative genes and improve the interpretability and visualization of tumor gene expression and single-cell sequencing data. Additionally, we developed a general framework named McbfsNW using gene expression data and protein interaction data to identify robust biomarkers and therapeutic targets for diagnosis and therapy of diseases. The framework incorporates the MCBFS algorithm, network recognition ensemble algorithm and feature selection wrapper. McbfsNW has been applied to the lung adenocarcinoma (LUAD) data sets. The preliminary results demonstrated that higher prediction results can be attained by identified biomarkers on the independent LUAD data set, and we also structured a drug-target network which may be good for LUAD therapy.ConclusionsThe proposed novel feature selection method is robust and effective for gene selection, classification, and visualization. The framework McbfsNW is practical and helpful for the identification of biomarkers and targets on genomic data. It is believed that the same methods and principles are extensible and applicable to other different kinds of data sets.

A consensus multi-view multi-objective gene selection approach for improved sample classification

BackgroundIn the field of computational biology, analyzing complex data helps to extract relevant biological information. Sample classification of gene expression data is one such popular bio-data analysis technique. However, the presence of a large number of irrelevant/redundant genes in expression data makes a sample classification algorithm working inefficiently. Feature selection is one such high-dimensionality reduction technique that helps to maximize the effectiveness of any sample classification algorithm. Recent advances in biotechnology have improved the biological data to include multi-modal or multiple views. Different ‘omics’ resources capture various equally important biological properties of entities. However, most of the existing feature selection methodologies are biased towards considering only one out of multiple biological resources. Consequently, some crucial aspects of available biological knowledge may get ignored, which could further improve feature selection efficiency.ResultsIn this present work, we have proposed a Consensus Multi-View Multi-objective Clustering-based feature selection algorithm called CMVMC. Three controlled genomic and proteomic resources like gene expression, Gene Ontology (GO), and protein-protein interaction network (PPIN) are utilized to build two independent views. The concept of multi-objective consensus clustering has been applied within our proposed gene selection method to satisfy both incorporated views. Gene expression data sets of Multiple tissues and Yeast from two different organisms (Homo Sapiens and Saccharomyces cerevisiae, respectively) are chosen for experimental purposes. As the end-product of CMVMC, a reduced set of relevant and non-redundant genes are found for each chosen data set. These genes finally participate in an effective sample classification.ConclusionsThe experimental study on chosen data sets shows that our proposed feature-selection method improves the sample classification accuracy and reduces the gene-space up to a significant level. In the case of Multiple Tissues data set, CMVMC reduces the number of genes (features) from 5565 to 41, with 92.73% of sample classification accuracy. For Yeast data set, the number of genes got reduced to 10 from 2884, with 95.84% sample classification accuracy. Two internal cluster validity indices - Silhouette and Davies-Bouldin (DB) and one external validity index Classification Accuracy (CA) are chosen for comparative study. Reported results are further validated through well-known biological significance test and visualization tool.

Clustering based feature selection using Partitioning Around Medoids (PAM)

High-dimensional data contains a large number of features. With many features, high dimensional data requires immense computational resources, including space and time. Several studies indicate that not all features of high dimensional data are relevant to classification result. Dimensionality reduction is inevitable and is required due to classifier performance improvement. Several dimensionality reduction techniques were carried out, including feature selection techniques and feature extraction techniques. Sequential forward feature selection and backward feature selection are feature selection using the greedy approach. The heuristics approach is also applied in feature selection, using the Genetic Algorithm, PSO, and Forest Optimization Algorithm. PCA is the most well-known feature extraction method. Besides, other methods such as multidimensional scaling and linear discriminant analysis. In this work, a different approach is applied to perform feature selection. Cluster analysis based feature selection using Partitioning Around Medoids (PAM) clustering is carried out. Our experiment results showed that classification accuracy gained when using feature vectors' medoids to represent the original dataset is high, above 80%.

Gene Expression Dataset Classification Using Artificial Neural Network and Clustering-Based Feature Selection

With the progression of bioinformatics, applications of GE profiles on cancer diagnosis along with classification have become an intriguing subject in the bioinformatics field. It holds numerous genes with few samples that make it arduous to examine and process. A novel strategy aimed at the classification of GE dataset as well as clustering-centered feature selection is proposed in the paper. The proposed technique first preprocesses the dataset using normalization, and later, feature selection was accomplished with the assistance of feature clustering support vector machine (FCSVM). It has two phases, gene clustering and gene representation. To make the chose top-positioned features worthy for classification, feature reduction is performed by utilizing SVM-recursive feature elimination (SVM-RFE) algorithm. Finally, the feature-reduced data set was classified using artificial neural network (ANN) classifier. When compared with some recent swarm intelligence feature reduction approach, FCSVM-ANN showed an elegant performance.

A Dynamic Two-Layers MI and Clustering-based Ensemble Feature Selection for Multi-Labels Text Classification

Multi-label text classification deals with the issue that arises from each sample being related to multiple labels. The text data suffers from high dimensionality. In order to resolve this issue, a feature selection (FS) method can be implemented for efficiently removing the noisy, irrelevant, and redundant features. Multi-label FS is a powerful tool for solving the high-dimension problem. With regards to handling correlation and high dimensionality problems in multi-label text classification, this paper investigates the various heterogeneous FS ensemble schemes. In addition, this paper proposes an enhanced FS method called dynamic multi-label two-layers MI and clustering-based ensemble feature selection algorithm (DMMC-EFS). The proposed method considers the: 1) dynamic global weight of feature, 2) heterogeneous ensemble, and 3) maximum dependency and relevancy and minimum redundancy of features. This method aims to overcome the high dimensionality of multi-label datasets and acquire improved multi-label text classification. We have conducted experiments based on three benchmark datasets: Reuters-21578, Bibtex, and Enron. The experimental results show that DMMC-EFS has significantly outperformed other state-of-the-art conventional and ensemble multi-label FS methods.

A network clustering based feature selection strategy for classifying autism spectrum disorder

BackgroundAdvanced non-invasive neuroimaging techniques offer new approaches to study functions and structures of human brains. Whole-brain functional networks obtained from resting state functional magnetic resonance imaging has been widely used to study brain diseases like autism spectrum disorder (ASD). Auto-classification of ASD has become an important issue. Existing classification methods for ASD are based on features extracted from the whole-brain functional networks, which may be not discriminant enough for good performance.MethodsIn this study, we propose a network clustering based feature selection strategy for classifying ASD. In our proposed method, we first apply symmetric non-negative matrix factorization to divide brain networks into four modules. Then we extract features from one of four modules called default mode network (DMN) and use them to train several classifiers for ASD classification.ResultsThe computational experiments show that our proposed method achieves better performances than those trained with features extracted from the whole brain network.ConclusionIt is a good strategy to train the classifiers for ASD based on features from the default mode subnetwork.

A cluster-based feature selection method for image texture classification

<p>Computer vision and pattern recognition applications have been counted serious research trends in engineering technology and scientific research content. These applications such as texture image analysis and its texture feature extraction. Several studies have been done to obtain accurate results in image feature extraction and classifications, but most of the extraction and classification studies have some shortcomings. Thus, it is substantial to amend the accuracy of the classification via minify the dimension of feature sets. In this paper, presents a cluster-based feature selection approach to adopt more discriminative subset texture features based on three different texture image datasets. Multi-step are conducted to implement the proposed approach. These steps involve texture feature extraction via Gray Level Co-occurrence Matrix (GLCM), Local Binary Pattern (LBP) and Gabor filter. The second step is feature selection by using K-means clustering algorithm based on five feature evaluation metrics which are infogain, Gain ratio, oneR, ReliefF, and symmetric. Finally, K-Nearest Neighbor (KNN), Naive Bayes (NB) and Support Vector Machine (SVM) classifiers are used to evaluate the proposed classification performance and accuracy. Research achieved better classification accuracy and performance using KNN and NB classifiers that were 99.9554% for Kelberg dataset and 99.0625% for SVM in Brodatz-1 and Brodatz-2 datasets consecutively. Conduct a comparison to other studies to give a unified view of the quality of the results and identify the future research directions.</p>