Ensemble Fuzzy Feature Selection Based on Relevancy, Redundancy, and Dependency Criteria

The main challenge of classification systems is the processing of undesirable data. Filter-based feature selection is an effective solution to improve the performance of classification systems by selecting the significant features and discarding the undesirable ones. The success of this solution depends on the extracted information from data characteristics. For this reason, many research theories have been introduced to extract different feature relations. Unfortunately, traditional feature selection methods estimate the feature significance based on either individually or dependency discriminative ability. This paper introduces a new ensemble feature selection, called fuzzy feature selection based on relevancy, redundancy, and dependency (FFS-RRD). The proposed method considers both individually and dependency discriminative ability to extract all possible feature relations. To evaluate the proposed method, experimental comparisons are conducted with eight state-of-the-art and conventional feature selection methods. Based on 13 benchmark datasets, the experimental results over four well-known classifiers show the outperformance of our proposed method in terms of classification performance and stability.

Feature Reduction and Optimization of Malware Detection System Using Ant Colony Optimization and Rough Sets

Malware is a malicious program that can cause a security breach of a system. Malware detection and classification is one of the burning topics of research in information security. Executable files are the major source of input for static malware detection. Machine learning techniques are very efficient in behavioral-based malware detection and need a dataset of malware with different features. In windows, malware can be detected by analyzing the portable executable (PE) files. This work contributes to identifying the minimum feature set for malware detection employing a rough set dependent feature significance combined with Ant Colony Optimization (ACO) as the heuristic-search technique. A malware dataset named claMP with both integrated features and raw features was considered as the benchmark dataset for this work. The analytical results prove that 97.15% and 92.8% data size optimization has been achieved with a minimum loss of accuracy for claMP integrated and raw datasets, respectively.

Identification of Key Features of CNS Drugs Based on SVM and Greedy Algorithm

Introduction: The research and development of drugs related to central nervous system (CNS) diseases is a long and arduous process with high cost, long cycle and low success rate. Identification of key features based on available CNS drugs is of great significance for the discovery of new drugs. Materials and Methods: In this paper, based on the PaDEL descriptors of CNS drugs and non-CNS drugs, a support vector machine (SVM) model was constructed to identify the key features of CNS drugs. Firstly, the random forest algorithm was used to rank descriptors according to the feature significance that contributes to the identification of CNS drugs. Then, a reliable SVM model was constructed, and the optimal combination of descriptors was determined based on greedy algorithm and recursive feature elimination method. Results and Conclusion: It was found, based on the optimal combination of 40 descriptors, the prediction accuracy of CNS drugs and non-CNS drugs reached 94.2% and 94.4% respectively. nF11HeteroRing, AATSC3v, SpMin6_Bhi, maxdssC, AATS4v, E1v, E3e, GATS5s, minsOH and minHBint4 are the key features to distinguish between CNS drugs and non-CNS drugs.

Feature Selection Ensemble

Many strategies have been exploited for the task of feature selection, in an effort to identify more compact and better quality feature subsets. Such techniques typically involve the use of an individual feature significance evaluation, or a measurement of feature subset consistency, that work together with a search algorithm in order to determine a quality subset. Feature selection ensemble aims to combine the outputs of multiple feature selectors, thereby producing a more robust result for the subsequent classifier learning tasks. In this paper, three novel implementations of the feature selection ensemble concept are introduced, generalising the ensemble approach so that it can be used in conjunction with many subset evaluation techniques, and search algorithms. A recently developed heuristic algorithm: harmony search is employed to demonstrate the approaches. Results of experimental comparative studies are reported in order to highlight the benefits of the present work. Thepaper ends with a proposal to extend the application of feature selection ensemble to aiding the development of biped robots (inspired by the authors’ involvement in the joint celebration of Olympic and the centenary of the birth of Alan Turing).