Paper Info
Title
Bacterial-inspired feature selection algorithm and its application in fault diagnosis of complex structures.
Abstract
Feature selection is an important preprocessing technique for data analysis and data mining. One of main challenge for feature selection is to overcome the curse of dimensionality. Bacterial algorithms, like Bacterial Foraging Optimization (BFO), have been well-exploited as the metaheuristics for addressing the optimization problems. In this paper, an extended bacterial algorithm named as Bacterial-Inspired Feature Selection Algorithm (BIFS) is proposed. In BIFS, the searching process of bacteria consists of two main mechanisms: interactive swimming (or running) strategy used in Bacterial Colony Optimization (BCO), and random tumbling strategy embedded in Bacterial Foraging Optimization (BFO). The rule controlled foraging mode in BCO has been used in BIFS to overcome the high computational cost problem in most BFOs. Meanwhile, the 'roulette wheel weighting' strategy is employed to weight the influence of features on the fitness functions and evaluate the distribution of the features within the large search space. Experiments on six benchmark datasets show that the proposed algorithm (i.e. BIFS) achieves higher classification accuracy rate in comparison to the four bacterial based algorithms and other three evolutionary algorithms. Furthermore, an additional real application of the proposed bacterial-inspired feature selection algorithm for fault diagnosis of complex structures in engineering has been developed. The results show that the proposed bacterial-inspired algorithm is capable of selecting the most sensitive sensors to detect and isolate the fault of complex structures.
Year
DOI
Venue
2016
10.1109/CEC.2016.7744272
CEC
Keywords
Field
DocType
GENE-EXPRESSION DATA,SUBSET-SELECTION,OPTIMIZATION
Data mining,Weighting,Evolutionary algorithm,Feature selection,Computer science,Artificial intelligence,Optimization problem,Metaheuristic,Mathematical optimization,Algorithm,Feature extraction,Curse of dimensionality,Statistical classification,Machine learning
Conference
Citations 
PageRank 
References 
0
0.34
9
Authors
3
Name
Order
Citations
PageRank
Hong Wang1639.27
X. J. Jing248534.25
Ben Niu323544.62