Variable-Number Differential Evolution and its Application on Uranium Ore Classification

Abstract:

Differential Evolution (DE) algorithm is a heuristic random search algorithm based on the group difference, and it is also an optimization algorithm which processes random search in the continuous space with actual number vector coding. In uranium ore classification, its impossible to know in advance the number of best features to select. This paper puts forward a new algorithm named Variable-number Differential Evolution (VDE) algorithm, and applies it to selecting the best features of hyperspectral data. Then applies k-nearest neighbor, decision tree, naïve bayes, naïve bayes tree and support vector machine algorithm on the newly obtained data set only containing the selected features, records average accuracy by 10-fold cross-validation. The results show that new algorithm can improve the accuracy of classification compared with genetic algorithm (GA) and original DE algorithm.

• Periodical: Advanced Materials Research (Volume 748)
• Pages: 640-645
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.748.640
• Online since: August 2013
• Authors: Chao Yu, Juan Yu, Zhi Hua Cai
• Keywords: Absorption Peaks, Classification, Differential Evolution, Feature Selection, Uranium Ore

References

[1] Ruiliang Pu, Peng Gong. Hyperspectral Remote Sensing and Its Applications. Beijing:Higher Education Press, 2001, pp.3-10.

[2] Thomas Villmann, Erzsebet Merenyi, Barbara Hammer. Neural Maps in Remote Sensing Image Analysis. Neural Networks. Vol. 16, 2003, pp.389-403.

DOI: 10.1016/s0893-6080(03)00021-2

[3] Rich Lawrence, Andrew Bunn, Scott Powell, Michael Zambon. Classification of remotely sensed imagery using stochastic gradient boosting as a refinement of classification tree analysis. Remote Sensing of Environment. Vol. 90(3), 2004, pp.331-336.

DOI: 10.1016/j.rse.2004.01.007

[4] Lihong Su, Mark J.Chopping, Albert Rango, John V.Martonchik, Debra P.C. Peters. Support Vector Machines for Recognition of Semi-arid Vegetation Types Using MISR Multi-angle Imagery. Remote Senesing of Environment. Vol. 112(5), 2008, pp.1927-1941.

DOI: 10.1016/j.rse.2006.05.023

[5] Shixin Yu, Steve De Backer, Paul Scheunders. Genetic Feature Selection Combined with Composite Fuzzy Nearest Neighbor classifiers for hyperspectral satellite imagery. Pattern Recognition Letters. Vol. 23, 2002. pp.183-190.

DOI: 10.1016/s0167-8655(01)00118-0

[6] Wenxia Chen, Ansheng Chen, Zhihua Cai. Research on Hyperspectral Absorption Feature Parameters Based Classification. Vol. 44, 2008, pp.230-241.

[7] Storn R, Price K. Differential Evolution-A Simple and Efficient Heurist for Global Optimization over Continuous Space. Joumal of Global Optimization. Vol. 11, 1997, pp.341-359.

[8] Lei Yu, Huan Liu. Feature Selection for High-Dimensional Data: A Fast Correlation-Based Filter Solution. In Proceedings of the Twentieth International Conference on Machine Learning. 2003, pp.856-863.

[9] Ian H.Witten, Eibe Frank. Data Mining: Practical Machine Learning Tools and Techniques. 2nd ed. San Francisco: Morgan Kaufmann Publications, 2005, pp.365-368.

DOI: 10.1186/1475-925x-5-51

[10] Jianguo He, Yuxian Mao, Jianzhong Li: Diagnostic spectral characteristics of damouritization in granite type uranium deposit. Uranium Geology. Vol. 24, 2008, pp.250-256