Abstract—In this paper, we propose apply ABC algorithm in analyzing microarray dataset. In addition, we propose an innovative hybrid classification model, Support Vector Machine (SVM) with ABC algorithm, to measure the classification accuracy for selected genes. We evaluate the performance of the proposed ABC-SVM algorithm by conducting extensive experiments on six binary and multi-class microarrays dataset. Furthermore, we compare our proposed ABC-SVM algorithm with previously known techniques. The experimental results prove that ABC-SVM algorithm is promising approach for solving gene selection and cancer classification problems, and achieves the highest classification accuracy together with the lowest average of selected genes compared to previously suggested methods. Index Terms—ABC, gene selection, microarray, and SVM. I INTRODUCTION Gene expression profiling or microarrays offer an efficient method of gathering data that can be used to determine the patterns of gene expression of many (if not all) genes in an organism in a single experiment. However, microarray dataset suffers from the curse of dimensionality, the small number of samples, and the level of irrelevant and noise genes, all of which makes the classification task for a given sample more challenging [1], [2]. Gene selection is the process of selecting the smallest subset of informative genes that are most predictive to its relative class using a classification model. This maximizes the classifier’ s ability to classify samples accurately. In this kind of study one aim is to identify the genes that contribute the most to cancer diagnosis, which would assist in drug discovery and early diagnosis. The optimal feature selection problem has been shown to be NP-hard [3]. Therefore, it is better to use heuristic approaches such as bio- inspired evolutionary algorithms in order to solve this problem. Bio-inspired evolutionary algorithms such as GA, PSO, and ABC are more applicable and accurate than the wrapper gene selection method [4] because they are capable of searching for optimal or near-optimal solutions on complex and large spaces of possible solutions. Furthermore, they allow searching the solution space by considering multiple interacting attributes simultaneously, rather than by Manuscript received February 26, 2016; revised June 6, 2016. The authors are with Computer Science Department, King Saud University, Saudi Arabia (e-mail: halshamlan@ksu.edu.sa, badrghada@hotmail.com, yousef@ksu.edu.sa). considering one attribute at a time [4]. The artificial bee colony (ABC) algorithm that is introduced by Karaboga [5] is one bio-inspired evolutionary approach that has been used to find an optimal solution in numerical optimization problems. The algorithm is inspired by the behavior of honey bees when seeking a quality food source. The performance of ABC algorithms has been compared with other evolutionary methods such as genetic algorithm (GA), differential evolution (DE), evolution strategies (ES), particle swarm optimization, and particle swarm-inspired evolutionary algorithm (PS-EA) [6]-[8]. Numerical comparison results showed that the ABC algorithm is competitive. Due to its simplicity and ease of implementation, the ABC algorithm has captured much attention and has been applied to solve many practical optimization problems. Therefore, in this paper, we propose the application of the ABC algorithm to select the predictive and informative genes from microarray gene expression profile. In this paper, we measure the efficiency of gene selection techniques using a support vector machine (SVM) as a classifier. An SVM displayed substantial benefits when compared to other classification approaches [9]. It is challenging to construct a linear classifier to separate the classes of data. An SVM addresses this problem by mapping the input space into a high-dimensional feature space; it then constructs a linear classification decision to classify the input data with a maximum margin hyperplane. An SVM has also been found to be more effective and faster than other machine learning methods, such as neural networks and k-nearest neighbour classifiers [10]. The proposed algorithm is tested using six binary and multi-class gene expression microarray datasets and is also compared with genetic algorithm when combined with SVM (GA-SVM), and particle swarm optimization when combined with SVM (PSO-SVM) algorithm. In addition, we compared it with other related algorithms that have been published recently. The experimental results show improvements in both the number of selected informative genes and cancer classification accuracy. The rest of this paper is organized as follows: general background about Artificial Bee Colony (ABC) algorithm will be presented in Section II. The proposed Artificial bee colony and SVM algorithms (ABC-SVM) for gene selection and cancer classification using Microarray dataset explained in Section III. Then, Section IV outlines the experimental setup and provide results. Finally, Section V concludes our paper. ABC-SVM: Artificial Bee Colony and SVM Method for Microarray Gene Selection and Multi Class Cancer Classification Hala M. Alshamlan, Ghada H. Badr, and Yousef A. Alohali International Journal of Machine Learning and Computing, Vol. 6, No. 3, June 2016 184 doi: 10.18178/ijmlc.2016.6.3.596