论文部分内容阅读
通过有效的特征选择进行疾病分类是当前生物信息学研究的常见问题,从高维基因数据中消除噪声、筛选出存在于低维子空间的特征基因,对预测、诊断和治疗疾病至关重要.基于新兴的联合稀疏优化算法与经典的独立法则,本文提出了联合稀疏独立分类方法.在特征选择时考虑了数据的整体稀疏结构及集体特征之间的关系,弥补了基因表达分析数据小样本信息不足的缺点.而利用独立法则进行疾病分类不仅方式简单、易于实现,并且有效阻断了噪声的相互干扰,具有较好的稳定性.在三个基因表达分析数据集上的疾病分类实验结果表明,新的分类方法具有良好的分类正确率和运行速度.
Disease classification through effective feature selection is a common problem in bioinformatics research. It is important to eliminate the noise from high-dimensional genetic data and screen out the characteristic genes existing in the low-dimensional subspace to predict, diagnose and treat diseases. Based on the emerging joint sparse optimization algorithm and classical independent law, this paper proposes a joint sparse independent classification method, which takes into account the relationship between the overall sparse structure of data and the collective features in feature selection and makes up for the small sample information of gene expression analysis data Deficiencies.And the use of independent laws for disease classification is not only simple, easy to implement, and effectively block the noise interference, with good stability in the three gene expression analysis data sets on the disease classification experiments show that , The new classification method has a good classification accuracy and speed.