Abstract

Feature selection is an important issue in the field of machine learning, which can reduce misleading computations and improve classification performance. Generally, feature selection can be considered as a binary optimization problem. Gravitational Search Algorithm (GSA) is a population-based heuristic algorithm inspired by Newton's laws of gravity and motion. Although GSA shows good performance in solving optimization problems, it has a shortcoming of premature convergence. In this paper, the concept of global memory is introduced and the definition of exponential Kbest is used in an improved version of GSA called IGSA. In this algorithm, the position of the optimal solution obtained so far is memorized, which can effectively prevent particles from gathering together and moving slowly. In this way, the exploitation ability of the algorithm gets improved, and a proper balance between exploration and exploitation gets established. Besides, the exponential Kbest can significantly decrease the running time. In order to solve feature selection problem, a binary IGSA (BIGSA) is further introduced. The proposed algorithm is tested on a set of standard datasets and compared with other algorithms. The experimental results confirm the high efficiency of BIGSA for feature selection.

Highlights

  • Machine learning has been widely employed in many technology and engineering fields, such as intrusion detection [1], pattern recognition [2], image analysis, text categorization [3], data mining [4] and multimedia information retrieval [5]

  • FEATURE SELECTION PROBLEM AND ITS SOLUTION a binary version of IGSA is further introduced, and a new feature selection technique based on the wrapper method is proposed, in which the K-Nearest Neighbor algorithm is used as a classifier to evaluate classification accuracy

  • In order to research the performance of binary IGSA (BIGSA) for feature selection, it is compared with the original algorithm and another modified algorithm: binary hybrid particle swarm optimization and gravitational search algorithm (BPSOGSA) [16], which is a classical modified version of Gravitational Search Algorithm (GSA)

Read more

Summary

INTRODUCTION

Machine learning has been widely employed in many technology and engineering fields, such as intrusion detection [1], pattern recognition [2], image analysis, text categorization [3], data mining [4] and multimedia information retrieval [5]. Wrapper methods use classifiers to evaluate candidate subsets obtained by search algorithms, and the feedback from classifiers is helpful for feature selection [8]. They guarantee classification accuracy through complex computation. In our previous work [17], a modified gravitational search algorithm is proposed, which improves the exploration ability of the algorithm and provides good performance for function optimization. Based on BIGSA, a new feature selection method is proposed, in which the K-Nearest Neighbor (K-NN) technique is used as a classifier to perform wrapper method and evaluate candidate subsets.

GRAVITATIONAL SEARCH ALGORITHM
FEATURE SELECTION PROBLEM AND ITS SOLUTION
FITNESS FUNCTION
1: Identify the feature dimension of dataset 2
K-NEAREST NEIGHBOR ALGORITHM
EXPERIMENTAL RESULTS
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call