تعيين ماشين¬هاي بردار پشتيبان بهينه در طبقه¬بندي تصاوير فرا طیفی بر مبناي الگوريتم ژنتيک
محورهای موضوعی : عمومى
فرهاد صمدزادگان
1
(دانشگاه تهران)
حديثه سادات حسني
2
(دانشگاه تهران)
کلید واژه: ماشین‌های بردار پشتيبان, تصاوير فرا طیفی, طبقه بندي, انتخاب مدل, انتخاب ويژگي, الگوريتم ژنتيک,
چکیده مقاله :
امروزه تصاوير فرا طیفی به علت غناي اطلاعات طيفي يک ابزار قوي و کارامد در سنجش از دور به حساب مي آيند و امکان تمايز بين عوارض مشابه را فراهم مي آورند. با توجه به پايداري ماشین‌های بردار پشتیبان در فضاهايي با ابعاد بالا، یک گزينه مناسب در طبقه بندي تصاوير فرا طیفی محسوب مي شوند. با اين وجود، عملکرد این طبقه بندي کننده ها تحت تأثیر پارامترها و فضاي ويژگي ورودي آن ها مي باشد. به منظور استفاده از ماشين هاي بردار پشتيبان با بيشترين کارایی، مي بايست مقادير بهينه ي پارامترها و همچنين زير مجموعه بهينه از ويژگي هاي ورودي تعيين گردند. در اين تحقيق از توانايي الگوريتم ژنتيک به عنوان يک تکنيک بهينه سازي فرا ابتکاري، در تعيين مقادير بهينه پارامترهاي ماشين هاي بردار پشتيبان و همچنين انتخاب زيرمجموعه ويژگي هاي بهينه در طبقه بندي تصاوير فرا طیفی استفاده شده است. نتايج عملي از به‌کارگیری روش فوق در خصوص داده هاي فرا طیفی سنجنده AVIRISنشان مي دهند، ويژگي هاي ورودي و پارامترها هر کدام جداگانه تأثیر بسزايي بر عملکرد ماشين هاي بردار پشتيبان دارند ولي بهترين عملکرد طبقه-بندي کننده با حل همزمان آن دو بدست مي آيد. در حل همزمان تعيين پارامتر و انتخاب ويژگي، براي کرنل گوسين و پلي نوميال به ترتيب 5% و 15% افزايش دقت با حذف بيش از نيمي از باندهاي تصوير حاصل شد. همچنين الگوريتم بهينه سازي شبيه سازي تبريد تدريجي به منظور مقايسه با الگوريتم ژنتيک پياده سازي شد که نتايج حاکي از برتري الگوريتم ژنتيک به ويژه با بزرگ و پيچيده شدن فضاي جستجو در رويکرد حل همزمان تعيين پارامتر و انتخاب ويژگي مي باشد.
۱٬۳۸۵ / ۵٬۰۰۰ Today, hyperspectral images are considered a powerful and efficient tool in remote sensing due to the wealth of spectral information and provide the possibility of distinguishing between similar complications. Considering the stability of support vector machines in spaces with high dimensions, they are considered a suitable option in the classification of hyperspectral images. Nevertheless, the performance of these classifiers is influenced by their input parameters and feature space. In order to use support vector machines with the highest efficiency, the optimal values of the parameters and also the optimal subset of the input features should be determined. In this research, the ability of the genetic algorithm as a meta-heuristic optimization technique has been used in determining the optimal values of support vector machine parameters and also selecting the subset of optimal features in the classification of hyperspectral images. The practical results of applying the above method on the hyperspectral data of AVIRIS sensor show that the input features and parameters each have a great effect on the performance of support vector machines, but the best performance of the classifier is obtained by solving them simultaneously. In the simultaneous solution of parameter determination and feature selection, for Gaussian kernel and polynomial, 5% and 15% increase in accuracy was achieved by removing more than half of the image bands. Also, the gradual cooling simulation optimization algorithm was implemented in order to compare with the genetic algorithm, and the results indicate the superiority of the genetic algorithm, especially with the large and complicated search space in the simultaneous solution approach of parameter determination and feature selection.
[1]. C. Chang, Hyperspectral data exploitation: theory and applications: Wiley-Blackwell, 2007.
[2]. G. Hughes, "On the mean accuracy of statistical pattern recognizers," IEEE Transactions on Information Theory, vol. 14, pp. 55-63, 2002.
[3]. G. Camps-Vallsand L. Bruzzone, "Kernel-based methods for hyperspectral image classification," IEEE Transactions on Geoscience and Remote Sensing, vol. 43, pp. 1351-1362, 2005.
[4]. P. Du, K. Tan, W. Zhang, and Z. Yan, "ANN Classification of OMIS Hyperspectral RemotelySensed Imagery: Experiments and Analysis," Congress on Image and Signal Processing, pp. 692-696, 2008.
[5]. T. Waheed, R. Bonnell, S. Prasher, and E. Paulet, "Measuring performance in precision agriculture: CART--A decision tree approach," Agricultural water management, vol. 84, pp. 173-185, 2006.
[6]. J. Ham, Y. Chen, M. Crawford, and J. Ghosh, "Investigation of the random forest framework for classification of hyperspectral data," IEEE Transactions on Geoscience and Remote Sensing, vol. 43, pp. 492-501, 2005.
[7]. F. Melgani and L. Bruzzone, "Classification of hyperspectral remote sensing images with support vector machines," IEEE Transactions on Geoscience and Remote Sensing, vol. 42, pp. 1778-1790, 2004.
[8]. C. Dai, X. Huang, and G. Dong, "Support Vector Machine for Classification of Hyperspectral Remote Sensing Imagery,"Fourth International Conference on Fuzzy System and Knowledge Discovery, pp. 77-80, 2007.
[9]. B. Guo, S. Gunn, R. Damper, and J. Nelson, "Customizing kernel functions for SVM-based hyperspectral image classification," IEEE Transactions onImage Processing, vol. 17, pp. 622-629, 2008.
[10]. P. Watanachaturaporn, M. Arora, and P. Varshney, "Hyperspectral image classification using support vector machines: A comparison with decision tree and neural network classifiers," 2006.
[11]. S. Arlot and A. Celisse, "A survey of cross-validation procedures for model selection," Statistics Surveys, vol. 4, pp. 40-79, 2010.
[12]. X. Zhang, X. Chen, and Z. He, "An ACO-based algorithm for parameter optimization of support vector machines," Expert systems with applications, 2010.
[13]. C. Wu, G. Tzeng, Y. Goo, and W. Fang, "A real-valued genetic algorithm to optimize the parameters of support vector machine for predicting bankruptcy," Expert systems with applications, vol. 32, pp. 397-408, 2007.
[14]. E. Huerta, B. Duval, and J. Hao, "A hybrid GA/SVM approach for gene selection and classification of microarray data," Applications of Evolutionary Computing, pp. 34-44, 2006.
[15]. L. Zhuo, J. Zheng, F. Wang, X. Li, B. Ai, and J. Qian, "A genetic algorithm based wrapper feature selection method for classification of hyperspectral images using support vector machine," The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. 37, pp. 397-402, 2008.
[16]. T. Zhang, X. Fu, R. Goh, C. Kwoh, and G. Lee, "A GA-SVM feature selection model based on high performance computing techniques," IEEE International Conference on System, Man and Cybernetics, pp. 2653-2658, 2009.
[17]. C. Huang, "ACO-based hybrid classification system with feature subset selection and model parameters optimization," Neurocomputing, vol. 73, pp. 438-448, 2009.
[18]. S. Lin, Z. Lee, S. Chen, and T. Tseng, "Parameter determination of support vector machine and feature selection using simulated annealing approach," Applied soft computing, vol. 8, pp. 1505-1512, 2008.
[19]. S. Lin, K. Ying, S. Chen, and Z. Lee, "Particle swarm optimization for parameter determination and feature selection of support vector machines," Expert systems with applications, vol. 35, pp. 1817-1824, 2008.
[20]. C. Huang and C. Wang, "A GA-based feature selection and parameters optimizationfor support vectormachines," Expert systems with applications, vol. 31, pp. 231-240, 2006.
[21]. R. Haupt, S. Haupt, and J. Wiley, Practical genetic algorithms: Wiley Online Library, 1998.
[22]. V. Vapnik, The nature of statistical learning theory: Springer Verlag, 2000.
[23]. A. Lorena and A. de Carvalho, "Evolutionary tuning of SVM parameter values in multiclass problems," Neurocomputing, vol. 71, pp. 3326-3334, 2008.
[24]. T. Weise, "Global Optimization Algorithms–Theory and Application,", Abrufdatum, vol. 1, 2008.
[25]. U. Maulik, "Medical image segmentation using genetic algorithms," IEEE Transactions onInformation Technology in Biomedicine, vol. 13, pp. 166-173, 2009.
[26]. E. Pourbasheer, S. Riahi, M. Ganjali, and P. Norouzi, "Application of genetic algorithm-support vector machine (GA-SVM) for prediction of BK-channels activity," European journal of medicinal chemistry, vol. 44, pp. 5023-5028, 2009.
[27]. B. de Souza, A. de Carvalho, R. Calvo, and R. Ishii, "Multiclass SVM model selection using particle swarm optimization," 2006, p. 31.
[28]. C. Hsu, C. Chang, and C. Lin, "A practical guide to support vector classification," Citeseer, 2003.
[29]. I. Guyon and A. Elisseeff, "An introduction to variable and feature selection," The Journal of Machine Learning Research, vol. 3, pp. 1157-1182, 2003.
[30]. M. Tahir, A. Bouridane, F. Kurugollu, and A. Amira, "Feature selection using tabu search for improving the classification rate of prostate needle biopsies," Pattern Recognition, vol. 2, pp. 335-33, 2004.
[31]. C. Tu, L. Chuang, J. Chang, and C. Yang, "Feature selection using PSO-SVM," IAENG International journal of computer science, vol. 33, pp. 111-116, 2007.
[32]. D. Niu, Y. Wang, and D. Wu, "Power load forecasting using support vector machine and ant colony optimization," Expert systems with applications, vol. 37, pp. 2531-2539, 2010.
[33]. H. Frohlich, O. Chapelle, and B. Scholkopf, "Feature selection for support vector machines by means of genetic algorithm," 2003, pp. 142-148.
[34]. S. Bhatia, P. Prakash, and G. Pillai, "SVM Based Decision Support System for Heart Disease Classification with Integer-Coded Genetic Algorithm to Select Critical Features," 2008.
[35]. A.B. Santos, C.S.F. de S. Celes, A. de A. Araŭjo, D. Menotti, "Feature selection for classification of remote sensed hyperspectral images: A filter approach using genetic algorithm and cluster validity,"The 2012 International Conference on Image Processing, Computer Vision, and Pattern Recognition (IPCV’12), 2012.
[36]. Y. Bazi and F. Melgani, "Toward an optimal SVM classification system for hyperspectral remote sensing images," IEEE Transactions onGeoscience and Remote Sensing, vol. 44, pp. 3374-3385, 2006.
[37]. I. Mejía-Guevara and Ákuri-Morales, "Evolutionary feature and parameter selection in support vector regression," MICAI 2007: Advances inArtificial Intelligence, pp. 399-408, 2007.
[38]. E. Avci, "Selecting of the optimal feature subset and kernel parameters in digital modulation classification by using hybrid genetic algorithm-support vector machines: HGASVM," Expert systems with applications, vol. 36, pp. 1391-1402, 2009.
[39]. B. F. de Souza and A. P. d. L. F. de Carvalho, "Gene selection based on multi-class support vector machines and genetic algorithms," Genetics and Molecular Research, pp. 599-607, 2005.
[40]. M. Zhao, C. Fu, L. Ji, K. Tang, and M. Zhou, "Feature selection and parameter determination for support vector machines: A new approach based on genetic algorithm with feature chromosomes," Expert System with Applications, Vol. 38, No. 5, pp. 5197-5204, 2011.
[41]. N. Chen, B. Ribeiro, A. S. Vieira, J. Duarte, J. C. Neves, "A genetic algorithm-based approach to cost-sensitive bankruptcy prediction," Expert System with Application, Vol. 38, No. 10, pp. 12939-12945, 2011.