【作者】 李菲；

【导师】 杨利英；

【作者基本信息】 西安电子科技大学 ， 计算机应用技术， 2013， 硕士

【摘要】 DNA微阵列数据广泛应用于生物医学，特别是癌症方面的研究。它是典型的高维小样本数据，具有高维，高冗余，高噪声的特点，许多经典的处理算法无法直接应用。线性判别分析(LDA)是模式识别的经典方法之一，在模式识别、数据分析等诸多领域取得了良好的应用效果。本文在分析和总结现有的LDA算法的基础上，将LDA应用于微阵列数据，取得了较好的结果。本文主要做了以下两方面工作：第一，分析总结了LDA算法在微阵列数据中的应用，提出了一种基于filter法则和ICA变换的LDA算法—FILDA算法。与传统的LDA算法处理微阵列数据相比，该算法消耗时间较少，且有效避免了矩阵变换奇异问题的产生，同时去除了特征中的冗余和噪音，提高了分类正确率。通过在Brain，Colondata，Prostate等微阵列数据集上的实验，以及同其他方法的对比实验，表明了本文算法的有效性。第二，由于独立分量分析具有计算不稳定性，本文借鉴了集成学习的思想设计了一种基于FILDA算法的集成系统，并进行了实验研究。实验结果表明，该系统在提高数据分类正确率的同时，具有较好的稳定性。DNA微阵列技术已经在医学以及生物学上得到越来越广泛的应用，其分类识别问题也是热点问题，本文即是对此的一个有益尝试。更多还原

【Abstract】 DNA microarray data is widely used in biomedical, especially cancer research.DNA microarray data is typical of the small sample of high-dimensional data, and hasthe characteristics of high-dimensional, high redundancy and has much noise. Manyclassic processing algorithms can not be applied directly. Linear discriminant analysis(LDA) is one of the classic pattern recognition method and make a good results inpattern recognition, data analysis, and many other areas. This paper will analyze andsummarize existing LDA algorithm and applied to microarray data to obtain betterresults based on LDA. This article is mainly to do the following two aspects:First, analysis the LDA algorithm in microarray data, proposed the LDAalgorithm-FILDA algorithm based on the filter law and ICA transform. Compared withthe traditional LDA algorithm processing microarray data, less time-consuming of thealgorithm, and effectively avoid the singularity problem of matrix transformationsproduce, eliminates redundancy and noise features to improve the classificationaccuracy. Through experiments on the the Brain, Colondata, Prostate microarraydatasets, as well as the comparative experiments with other methods, prove theeffectiveness of the proposed algorithm.Second, due to the independent component analysis has calculated instability, thispaper draws on the idea of ensenmble learning and designs based on the the FILDAalgorithm. The experiments show that the system improves the data classificationaccuracy and has good stability at the same time.DNA microarray technology has been more widely used in medicine and biology,its classification and recognition is also a hot issue。This article is a useful attempt.更多还原