Threshold Controlled Binary Particle Swarm Optimization for High Dimensional Feature Selection

Abstract

Dimensionality reduction or the optimal selection of features is a challenging task due to large search space. Currently, many research has been performed in this domain to improve the accuracy as well as to minimize the computational complexity. Particle Swarm Optimization (PSO) based feature selection approach seems very promising and has been extensively used for this work. In this paper, a Threshold Controlled Binary Particle Swarm Optimization (TC-BPSO) along with Multi-Class Support Vector Machine (MC-SVM) is proposed and compared with Conventional Binary Particle Swarm Optimization (C-BPSO). TC-BPSO is used for the selection of features while MC-SVM is used to calculate the classification accuracy. 70% of the data is used to train the MC-SVM model while the test has been performed on rest 30% data to calculate the accuracy. Proposed approach is tested on ten different datasets having varying difficulties such as some datasets having large number of features while some have small, some have just two classes while some have many classes, some datasets having small number of instances while some have large number of instances and the results obtained on these datasets are compared with some of the existing methods. Experiments show that the obtained results are very promising and achieved the best accuracy in minimum possible features. Proposed approach outperforms C-BPSO in all contexts on most of the datasets and 3-4 times computationally faster. It also outperforms in all context when compared with the existing work and 5-8 times computationally faster.

Edge Detection using Distinct Particle Swarm Optimization

“…There are numerous edge detection calculations that endeavor to manage noise\commotion, yet here we just layout the "Canny edge detector", it used for examination with the new calculation. "The "Canny" edge detector, as a "Gaussian" channel based calculation, decides the edges of an image dependent on an optimization procedure to locate a maximal of the slope size of an image which is flatten by the Gaussian channel [12].This calculation is exceptionally well known in light of the fact that it is a total procedure of edge detection and has great localization. Ordinary strides of the "Canny" edge detector" are as per the following: the noise from the image are removed by means of filtering; figuring the inclination greatness and heading for every pixel in the image; "utilizing "non-maxima suppression (NMS)" calculation to smother non-maxima edges through which there is no pixel among its neighbors in the slope course with bigger angle extent; and distinguishing the edges and connecting the broken edges (using a procedure, for example, a hysteresis thresholding method [11])"…”

“…"In PSO technique, there is a population of particles or individuals and every particle has a finite memory space to keep the records of past states. There are number of applications where PSO has been used such as training neural networks [11], optimizing power systems [14], fuzzy control systems [14], robotics [12] etc. " In "PSO" technique there is numbers of n particles having the population which progress along an m-dimensional search space.…”

“…"When the foremost curve is detected by the "PSO algorithm", the pixels comes under the curve are treated as edges and the pixels inside the figure with red line are not treated as processed pixels; if not all pixels inside the figure with red line are treated as processed pixels. The processed pixels are taken into consideration in the next runs because another curve may be found in this area [11,12]. "…”

“…The cost of deflection(IC) of an pixel of an edge which gives local calculation of deflection is based on the motion of direction from pixel to its next neighboring pixel on the unbroken edge [14,15]. (12) where is the direction of movement encoded by a particle and 3 = 40, which is selected by experiments. The cost of inflection of the curve given by each individual can be calculated with the help of Equation (12)."…”

“…(12) where is the direction of movement encoded by a particle and 3 = 40, which is selected by experiments. The cost of inflection of the curve given by each individual can be calculated with the help of Equation (12)." (13) 4) "New Objective Function:" The proposed objective function can be improved with the help of "PSO" technique to detect the edges in an image more smoothly and accurately.…”