基于Isomap算法的地震属性参数降维处理
|
摘要
针对非线性高维地震属性参数降维的难题,在综合分析研究工区地震属性参数数据实际特征的基础上,引入非线性降维Isomap算法,并基于MATLAB平台进行了算法的程序编制,进一步将Isomap降维的结果与线性的MDS降维结果通过小波神经网络进行检验,从算法原理的角度讨论了Isomap算法在地震属性参数降维处理中的可适性,表明Isomap具有更强的降维能力和发现数据本质结构的能力,从另一个角度提供了解决地质数据处理问题的方法。
In view of difficulty in reduction dimension of nonlinear high dimensional seismic attribution parameter and based on comprehensive analysis of the exact character of seismic attribution parameter in the study area, a method of nonlinear Isomap algorithm is introduced. The algorithm program is achieved on the base of MATLAB. Further, the result from Isomap and that from linear MDS are examined by wavelet neural net. Whether Isomap is applicable for reduction dimension of seismic attribution parameter or not is discussed from the angle of algorithm theory. It is shown that Isomap has stronger ability in reduction dimension and finding data's essential structure. It provides a solution to analyze geological data from another angle.
In view of difficulty in reduction dimension of nonlinear high dimensional seismic attribution parameter and based on comprehensive analysis of the exact character of seismic attribution parameter in the study area, a method of nonlinear Isomap algorithm is introduced. The algorithm program is achieved on the base of MATLAB. Further, the result from Isomap and that from linear MDS are examined by wavelet neural net. Whether Isomap is applicable for reduction dimension of seismic attribution parameter or not is discussed from the angle of algorithm theory. It is shown that Isomap has stronger ability in reduction dimension and finding data's essential structure. It provides a solution to analyze geological data from another angle.
引文
[1]JOSHUA B TENENBAUM,VIN DE SILVA,LANG-FORDJ C.A global geometric framework for nonlinear di mensionality reduction[J].Science,2000(290):2319-2323.
[2]ROWEIS S T,SAUL L K.Nonlinear di mensionality re-duction by locally linear embedding[J].Science,2000,290:2323-2326.
[3]ZHANG ZHENGYUE,ZHA HONGYUAN.Principal manifolds and nonlinear di mensinality reduction via local tangent space alignment[R].CSE-02-019,Technical Re-port,CSE,Penn State Univ,2002.
[4]SHIFENG WENG,CHANGSHUI ZHANG,ZHONG-LIN LIN.Exploring the structure of supervised data by discri minant isometric mapping[J].Pattern Recognition,2005,38:599-601.
[5]LEE J A,LENDASSE J A,VERLEYSEN M.Nonlinear projection with curvilinear distances:Isomap versus cur-vilinear distance analysis[J].Neurocomputing,2004,57(1):49-76.
[6]HONG TANG,TAO FANG,PENG-FEI SHI.Nonlinear discri minant mapping using the laplacian of a graph[J].Pattern Recognition,2006,39(1):156-159.
[7]徐志节,杨杰.利用非线性降维方法预测膜蛋白类型[J].上海交通大学学报,2005,39(2):279-281.
[8]赵连伟,罗四维,赵艳敞,等.高维数据流形的低维嵌入及维数研究[J].软件学报,2005,16(8):1423-1430.
[9]侯越先,丁峥,何丕廉.基于自组织的鲁棒非线性维数约减算法[J].计算机研究与发展,2005,42(2):188-195.
[10]魏艳,尹成,丁峰,等.地震多属性综合分析的应用研究[J].石油物探,2007,46(1):42-47.
[2]ROWEIS S T,SAUL L K.Nonlinear di mensionality re-duction by locally linear embedding[J].Science,2000,290:2323-2326.
[3]ZHANG ZHENGYUE,ZHA HONGYUAN.Principal manifolds and nonlinear di mensinality reduction via local tangent space alignment[R].CSE-02-019,Technical Re-port,CSE,Penn State Univ,2002.
[4]SHIFENG WENG,CHANGSHUI ZHANG,ZHONG-LIN LIN.Exploring the structure of supervised data by discri minant isometric mapping[J].Pattern Recognition,2005,38:599-601.
[5]LEE J A,LENDASSE J A,VERLEYSEN M.Nonlinear projection with curvilinear distances:Isomap versus cur-vilinear distance analysis[J].Neurocomputing,2004,57(1):49-76.
[6]HONG TANG,TAO FANG,PENG-FEI SHI.Nonlinear discri minant mapping using the laplacian of a graph[J].Pattern Recognition,2006,39(1):156-159.
[7]徐志节,杨杰.利用非线性降维方法预测膜蛋白类型[J].上海交通大学学报,2005,39(2):279-281.
[8]赵连伟,罗四维,赵艳敞,等.高维数据流形的低维嵌入及维数研究[J].软件学报,2005,16(8):1423-1430.
[9]侯越先,丁峥,何丕廉.基于自组织的鲁棒非线性维数约减算法[J].计算机研究与发展,2005,42(2):188-195.
[10]魏艳,尹成,丁峰,等.地震多属性综合分析的应用研究[J].石油物探,2007,46(1):42-47.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心