摘要
在数据分类算法的实际应用中,经常会遇到数据不平衡的问题(即正负样本的数目相差极大)。标准的分类算法在处理这一问题时,往往很难达到令人满意的性能。提出一种新的方法,通过对正负样本分别进行核函数拟合,根据拟合好的核函数对未知样本进行预测。在UCI标准数据集的仿真实验结果表明,该方法能有效地处理非平衡数据问题。
The data imbalance problem,where the difference between numbers of positive samples and negative samples are much great,frequently occurs in practical application of data classification algorithms.Standard classification algorithm is hard to perform satisfactorily when dealing with this.In this paper,the author propose a new method,which fits the positive samples and negative samples respectively with kernel functions and predicts the unknown samples according to the fitted kernel function.Simulation experiment results made on UCI standard data sets shown that the proposed method can effectively deal with the imbalanced data problems.
出处
《计算机应用与软件》
CSCD
2010年第4期177-179,共3页
Computer Applications and Software
关键词
非平衡数据
核函数
拟合
Imbalanced data Kernel function Fitting