quote
LaTex
@inproceedings{Xue:2014:DEM:2598394.2598493,
author = {Xue, Bing and Fu, Wenlong and Zhang, Mengjie},
title = {Differential Evolution (DE) for Multi-objective Feature Selection in Classification},
booktitle = {Proceedings of the Companion Publication of the 2014 Annual Conference on Genetic and Evolutionary Computation},
series = {GECCO Comp ‘14},
year = {2014},
isbn = {978-1-4503-2881-4},
location = {Vancouver, BC, Canada},
pages = {83–84},
numpages = {2},
url = {http://doi.acm.org/10.1145/2598394.2598493},
doi = {10.1145/2598394.2598493},
acmid = {2598493},
publisher = {ACM},
address = {New York, NY, USA},
keywords = {classification, differential evolution, feature selection, multi-objective optimisation},
}
Normal
Bing Xue, Wenlong Fu, and Mengjie Zhang. 2014. Differential evolution (DE) for multi-objective feature selection in classification. In Proceedings of the Companion Publication of the 2014 Annual Conference on Genetic and Evolutionary Computation (GECCO Comp ‘14). ACM, New York, NY, USA, 83-84. DOI=http://dx.doi.org/10.1145/2598394.2598493
Summary
- minimise the number of features
- maximise the classification accuracy
Evolutionary computation techniques
mult-objective tasks
differential evolution (DE)
multi-objective feature selection algorithm (DEMOFS)
main content
Single Objective Algorithms: DEFS
Real-valued encoding
of length — number of features
DE / rand / 1 / bin
the training set
the test set
Multi-objective Algorithm: DEMOFS
a multi-objective DE algorithm — DEMO
based on non-dominated sorting based genetic algorithm II (NSGAII)
test
70% as the training set and 30% as the test set
30 independent runs on each dataset
two traditional feature selection algorithms (LFS and GSBS)
nine commonly used datasets