A multi-objective feature selection method based on bacterial foraging optimization
- PDF / 722,415 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 9 Downloads / 190 Views
(0123456789().,-volV)(0123456789(). ,- volV)
A multi-objective feature selection method based on bacterial foraging optimization Ben Niu1 • Wenjie Yi1 • Lijing Tan1 • Shuang Geng1
•
Hong Wang1
Ó Springer Nature B.V. 2019
Abstract Feature selection plays an important role in data preprocessing. The aim of feature selection is to recognize and remove redundant or irrelevant features. The key issue is to use as few features as possible to achieve the lowest classification error rate. This paper formulates feature selection as a multi-objective problem. In order to address feature selection problem, this paper uses the multi-objective bacterial foraging optimization algorithm to select the feature subsets and k-nearest neighbor algorithm as the evaluation algorithm. The wheel roulette mechanism is further introduced to remove duplicated features. Four information exchange mechanisms are integrated into the bacteria-inspired algorithm to avoid the individuals getting trapped into the local optima so as to achieve better results in solving high-dimensional feature selection problem. On six small datasets and ten high-dimensional datasets, comparative experiments with different conventional wrapper methods and several evolutionary algorithms demonstrate the superiority of the proposed bacteria-inspired based feature selection method. Keywords Feature selection Multi-objective optimization Bacterial foraging optimization Information exchange mechanism
1 Introduction With the development of information technology, the main challenge of data mining now is how to extract useful feature information from existing enormous data rather than how to collect a large amount of data. Feature selection can eliminate irrelevant or redundant features, so that it assists in reducing the number of features, improving model accuracy, and shortening running time. On the other hand, selecting truly relevant features can simplify the model, making it easier for researchers to understand how data is produced. Many researchers proposed various methods to select the most suitable features. Some previous
Shuang Geng and Hong Wang contributed equally to this article. & Shuang Geng [email protected] & Hong Wang [email protected] 1
College of Management, Shenzhen University, Shenzhen 518060, China
researches have viewed feature selection as a single-objective problem to minimize the classification error rate. Actually, feature selection problem can also be regarded as selecting a feature subset from an original set with the minimum feature subset size (Hamdani et al. 2007). This problem can be defined as a multi-objective problem. Traditional feature selection algorithms can be classified into mainly two types: filter method and wrapper method (Jovic´ and Bogunovic´ 2015). The main principle of filter method is to use evaluation criterions to enhance correlation between individual features and classes, and to reduce correlation among features simultaneously. Filter methods save the training steps of classifier, as a result, the
Data Loading...
