Feature Selection with Binary Symbiotic Organisms Search Algorithm for Email Spam Detection

Abstract

One method to increase classifier accuracy is using Feature Selection (FS). The main idea in the FS is reducing complexity, eliminating irrelevant information, and deleting a subset of input features that either have little information or have no information for prediction. In this paper, three efficient binary methods based on the Symbiotic Organisms Search (SOS) algorithm were presented for solving the FS problem. In the first and second methods, several S_shaped and V_shaped transfer functions were used for the binarization of the SOS, respectively. These methods were called BSOSS and BSOSV. In the third method, two new operators called Binary Mutualism Phase (BMP) and Binary Commensalism Phase (BCP) were presented for binarization of the SOS, named Efficient Binary SOS (EBSOS). The proposed methods were run on 18 standard UCI datasets and compared to the base and important meta-heuristic algorithms. The test results showed that the EBSOS method has the best performance among the three proposed methods for the binarization of the SOS. Finally, the EBSOS method was compared to the Genetic Algorithm (GA), Binary Bat Algorithm (BBA), Binary Particle Swarm Optimization (BPSO) Algorithm, Binary Flower Pollination Algorithm (BFPA), Binary Grey Wolf Optimizer (BGWO) Algorithm, Binary Dragonfly Algorithm (BDA), and Binary Chaotic Crow Search Algorithm (BCCSA). In addition, the EBSOS method was executed on the spam email dataset with the KNN, NB, SVM, and MLP classifiers. The results showed that the EBSOS method has better performance compared to other methods in terms of feature count and accuracy criteria. Furthermore, it was practically evaluated on spam email detection in particular.

EWOA-OPF: Effective Whale Optimization Algorithm to Solve Optimal Power Flow Problem

“…Traditional optimization methods have several drawbacks when solving complex and complicated problems that require considerable time and cost optimization. Metaheuristic algorithms have been proven capable of handling a variety of continuous and discrete optimization problems [46] in a wide range of applications including engineering [47][48][49], industry [50,51], image processing and segmentation [52][53][54], scheduling [55,56], photovoltaic modeling [57,58], optimal power flow [59,60], power and energy management [61,62], planning and routing problems [63][64][65], intrusion detection [66,67], feature selection [68][69][70][71][72], spam detection [73,74], medical diagnosis [75][76][77], quality monitoring [78], community detection [79], and global optimization [80][81][82]. In the following, some representative metaheuristic algorithms from the swarm intelligence category used in our experiments are described.…”

Hybrid leader based optimization: a new stochastic optimization algorithm for solving optimization applications

“…Scientists’ research in optimization studies also includes improving existing algorithms 44 – 47 , extending hybrid algorithms by combining different algorithms to increase their efficiency 48 , 49 , and developing binary versions of optimization algorithms 50 – 53 .…”

MTV-MFO: Multi-Trial Vector-Based Moth-Flame Optimization Algorithm

“…Therefore, it is still necessary to propose new algorithms or improve existing algorithms to eliminate their defects. These algorithms have been used to solve several real-world problems in both continuous and discrete spaces such as feature selection [58][59][60][61], scheduling and planning [62], disease diagnosis [63], engineering problems [64], photovoltaic energy generation systems [65,66], economic dispatch problems [67], global optimization [68][69][70], community detection [71][72][73], and motion estimation [74,75]. Among swarm intelligence algorithms, the moth flame optimization (MFO) algorithm has attracted noticeable interest for optimization purposes.…”