On fuzzy-rough sets approach to feature selection

Covering assisted intuitionistic fuzzy bi-selection technique for data reduction and its applications

The dimension and size of data is growing rapidly with the extensive applications of computer science and lab based engineering in daily life. Due to availability of vagueness, later uncertainty, redundancy, irrelevancy, and noise, which imposes concerns in building effective learning models. Fuzzy rough set and its extensions have been applied to deal with these issues by various data reduction approaches. However, construction of a model that can cope with all these issues simultaneously is always a challenging task. None of the studies till date has addressed all these issues simultaneously. This paper investigates a method based on the notions of intuitionistic fuzzy (IF) and rough sets to avoid these obstacles simultaneously by putting forward an interesting data reduction technique. To accomplish this task, firstly, a novel IF similarity relation is addressed. Secondly, we establish an IF rough set model on the basis of this similarity relation. Thirdly, an IF granular structure is presented by using the established similarity relation and the lower approximation. Next, the mathematical theorems are used to validate the proposed notions. Then, the importance-degree of the IF granules is employed for redundant size elimination. Further, significance-degree-preserved dimensionality reduction is discussed. Hence, simultaneous instance and feature selection for large volume of high-dimensional datasets can be performed to eliminate redundancy and irrelevancy in both dimension and size, where vagueness and later uncertainty are handled with rough and IF sets respectively, whilst noise is tackled with IF granular structure. Thereafter, a comprehensive experiment is carried out over the benchmark datasets to demonstrate the effectiveness of simultaneous feature and data point selection methods. Finally, our proposed methodology aided framework is discussed to enhance the regression performance for IC50 of Antiviral Peptides.

Adaptive neighborhood rough set model for hybrid data processing: a case study on Parkinson's disease behavioral analysis

Extracting knowledge from hybrid data, comprising both categorical and numerical data, poses significant challenges due to the inherent difficulty in preserving information and practical meanings during the conversion process. To address this challenge, hybrid data processing methods, combining complementary rough sets, have emerged as a promising approach for handling uncertainty. However, selecting an appropriate model and effectively utilizing it in data mining requires a thorough qualitative and quantitative comparison of existing hybrid data processing models. This research aims to contribute to the analysis of hybrid data processing models based on neighborhood rough sets by investigating the inherent relationships among these models. We propose a generic neighborhood rough set-based hybrid model specifically designed for processing hybrid data, thereby enhancing the efficacy of the data mining process without resorting to discretization and avoiding information loss or practical meaning degradation in datasets. The proposed scheme dynamically adapts the threshold value for the neighborhood approximation space according to the characteristics of the given datasets, ensuring optimal performance without sacrificing accuracy. To evaluate the effectiveness of the proposed scheme, we develop a testbed tailored for Parkinson's patients, a domain where hybrid data processing is particularly relevant. The experimental results demonstrate that the proposed scheme consistently outperforms existing schemes in adaptively handling both numerical and categorical data, achieving an impressive accuracy of 95% on the Parkinson's dataset. Overall, this research contributes to advancing hybrid data processing techniques by providing a robust and adaptive solution that addresses the challenges associated with handling hybrid data, particularly in the context of Parkinson's disease analysis.

Fuzzy Multi-objective Programming Approach for Constrained Matrix Games with Payoffs of Fuzzy Rough Numbers

Imprecise constrained matrix games (such as fuzzy constrained matrix games, interval-valued constrained matrix games, and rough constrained matrix games) have attracted considerable research interest. This article is concerned with developing an effective fuzzy multi-objective programming algorithm to solve constraint matrix games with payoffs of fuzzy rough numbers (FRNs). For simplicity, we refer to this problem as fuzzy rough constrained matrix games. To the best of our knowledge, there are no previous studies that solve the fuzzy rough constrained matrix games. In the proposed algorithm, it is proven that a constrained matrix game with fuzzy rough payoffs has a fuzzy rough-type game value. Moreover, this article constructs four multi-objective linear programming problems. These problems are used to obtain the lower and upper bounds of the fuzzy rough game value and the corresponding optimal strategies of each player in any fuzzy rough constrained matrix games. Finally, a real example of the market share game problem demonstrates the effectiveness and reasonableness of the proposed algorithm. Additionally, the results of the numerical example are compared with the GAMS software results. The significant contribution of this article is that it deals with constraint matrix games using two types of uncertainties, and, thus, the process of decision-making is more flexible.

Combining Fuzzy C-Means Clustering with Fuzzy Rough Feature Selection

With the rapid development of the network, data fusion becomes an important research hotspot. Large amounts of data need to be preprocessed in data fusion; in practice, the features of datasets can be filtered to reduce the amount of data. The feature selection based on fuzzy rough sets can process a large number of continuous and discrete data to reduce the data dimension, making the selected feature subset highly correlated with the classification but less dependent on other features. In this paper, a new method of fuzzy rough feature selection is proposed which combines the membership function determination method of fuzzy c-means clustering and fuzzy equivalence to the original selection. Different from the existing research, our method takes full advantage of knowledge about the dataset itself and the differences between datasets, which makes the features selected have a higher correlation with the classification, improves the classification accuracy, and reduces the data dimension. Experimental results on the UCI machine learning repository datasets confirmed the performance and effectiveness of our method. Compared to the existing method, smaller subsets of features and an average of 1% higher classification accuracies were achieved.

Prediction of cancer using customised fuzzy rough machine learning approaches

This Letter proposes a customised approach for attribute selection applied to the fuzzy rough quick reduct algorithm. The unbalanced data is balanced using synthetic minority oversampling technique. The huge dimensionality of the cancer data is reduced using a correlation-based filter. The dimensionality reduced balanced attribute gene subset is used to compute the final minimal reduct set using a customised fuzzy triangular norm operator on the fuzzy rough quick reduct algorithm. The customised fuzzy triangular norm operator is used with a Lukasiewicz fuzzy implicator to compute the fuzzy approximation. The customised operator selects the least number of informative feature genes from the dimensionality reduced datasets. Classification accuracy using leave-one-out cross validation of 94.85, 76.54, 98.11, and 99.13% is obtained using a customised function for Lukasiewicz triangular norm operator on leukemia, central nervous system, lung, and ovarian datasets, respectively. Performance analysis of the conventional fuzzy rough quick reduct and the proposed method are performed using parameters such as classification accuracy, precision, recall, F-measure, scatter plots, receiver operating characteristic area, McNemar test, chi-squared test, Matthew's correlation coefficient and false discovery rate that are used to prove that the proposed approach performs better than available methods in the literature.

About a Fuzzy Distance between Two Fuzzy Partitions and Application in Attribute Reduction Problem

According to traditional rough set theory approach, attribute reduction methods are performed on the decision tables with the discretized value domain, which are decision tables obtained by discretized data methods. In recent years, researches have proposed methods based on fuzzy rough set approach to solve the problem of attribute reduction in decision tables with numerical value domain. In this paper, we proposeafuzzy distance between two partitions and an attribute reduction method in numerical decision tables based on proposed fuzzy distance. Experiments on data sets show that the classification accuracy of proposed method is more efficient than the ones based fuzzy entropy.

Feature selection inspired classifier ensemble reduction

Classifier ensembles constitute one of the main research directions in machine learning and data mining. The use of multiple classifiers generally allows better predictive performance than that achievable with a single model. Several approaches exist in the literature that provide means to construct and aggregate such ensembles. However, these ensemble systems contain redundant members that, if removed, may further increase group diversity and produce better results. Smaller ensembles also relax the memory and storage requirements, reducing system's run-time overhead while improving overall efficiency. This paper extends the ideas developed for feature selection problems to support classifier ensemble reduction, by transforming ensemble predictions into training samples, and treating classifiers as features. Also, the global heuristic harmony search is used to select a reduced subset of such artificial features, while attempting to maximize the feature subset evaluation. The resulting technique is systematically evaluated using high dimensional and large sized benchmark datasets, showing a superior classification performance against both original, unreduced ensembles, and randomly formed subsets.

Leafcutter Ant Colony Optimization Algorithm for Feature Subset Selection on Classifying Digital Mammograms

Ant Colony Optimization (ACO) has been applied in wide range of applications. In ACO, for every iteration the entire problem space is considered for the solution construction using the probability of the pheromone deposits. After convergence, the global solution is made with the path which has highest pheromone deposit. In this paper, a novel solution construction technique has been proposed to reduce the time complexity and to improve the performance of the ACO. The idea is derived from the behavior of a special ant species called ‘Leafcutter Ants', they spend much of their time for cutting leaves to make fertilizer to gardens in which they grow the fungi that they eat. This behavior is incorporated with the general ACO algorithm to propose a novel feature selection method called ‘Leafcutter Ant Colony Optimization' (LACO) algorithm. The LACO has been applied to select the relevant features for digital mammograms and their corresponding classification performance is studied and compared.

Dimensionality reduction based on fuzzy rough sets oriented to ischemia detection

This paper presents a dimensionality reduction study based on fuzzy rough sets with the aim of increasing the discriminant capability of the representation of normal ECG beats and those that contain ischemic events. A novel procedure is proposed to obtain the fuzzy equivalence classes based on entropy and neighborhood techniques and a modification of the Quick Reduct Algorithm is used to select the relevant features from a large feature space by a dependency function. The tests were carried out on a feature space made up by 840 wavelet features extracted from 900 ECG normal beats and 900 ECG beats with evidence of ischemia. Results of around 99% classification accuracy are obtained. This methodology provides a reduced feature space with low complexity and high representation capability. Additionally, the discriminant strength of entropy in terms of representing ischemic disorders from time-frequency information in ECG signals is highlighted.

Seeding and harvest: a framework for unsupervised feature selection problems

Feature selection, also known as attribute selection, is the technique of selecting a subset of relevant features for building robust object models. It is becoming more and more important for large-scale sensors applications with AI capabilities. The core idea of this paper is derived from a straightforward and intuitive principle saying that, if a feature subset (pattern) has more representativeness, it should be more self-organized, and as a result it should be more insensitive to artificially seeded noise points. In the light of this heuristic finding, we established the whole set of theoretical principles, based on which we proposed a two-stage framework to evaluate the relative importance of feature subsets, called seeding and harvest (S&H for short). At the first stage, we inject a number of artificial noise points into the original dataset; then at the second stage, we resort to an outlier detector to identify them under various feature patterns. The more precisely the seeded points can be extracted under a particular feature pattern, the more valuable and important the corresponding feature pattern should be. Besides, we compared our method with several state-of-the-art feature selection methods on a number of real-life datasets. The experiment results significantly confirm that our method can accomplish feature reduction tasks with high accuracy as well as low computing complexity.

Prediction of protein interaction hot spots using rough set-based multiple criteria linear programming

Protein-protein interactions are fundamentally important in many biological processes and it is in pressing need to understand the principles of protein-protein interactions. Mutagenesis studies have found that only a small fraction of surface residues, known as hot spots, are responsible for the physical binding in protein complexes. However, revealing hot spots by mutagenesis experiments are usually time consuming and expensive. In order to complement the experimental efforts, we propose a new computational approach in this paper to predict hot spots. Our method, Rough Set-based Multiple Criteria Linear Programming (RS-MCLP), integrates rough sets theory and multiple criteria linear programming to choose dominant features and computationally predict hot spots. Our approach is benchmarked by a dataset of 904 alanine-mutated residues and the results show that our RS-MCLP method performs better than other methods, e.g., MCLP, Decision Tree, Bayes Net, and the existing HotSprint database. In addition, we reveal several biological insights based on our analysis. We find that four features (the change of accessible surface area, percentage of the change of accessible surface area, size of a residue, and atomic contacts) are critical in predicting hot spots. Furthermore, we find that three residues (Tyr, Trp, and Phe) are abundant in hot spots through analyzing the distribution of amino acids.