摘要
针对训练模式类标签不精确的识别问题,提出了基于可传递信度模型(TBM)的自适应k-NN分类器,它通过运用pignistic变换,可以方便地对待识别模式真正所属的类做出决策,并通过梯度下降来最小化训练模式的输出类标签与目标类标签之间的误差函数,以实现参数的自适应学习.实验表明,该分类器用于处理训练模式类标签不精确的模式识别问题是有效的,且与参数优化前的基于TBM的k-NN分类器相比,其误分类率更低、鲁棒性更强.
For processing training patterns with imprecise class labels, an adaptive fuzzy k-nearest neighbor classifier based on the transferable belief model(TBM) is presented in the paper. It's convenient to make decision about the true class membership of a pattern to be classified through the application of the pignistic transformation. And the parameters in the classifier are tuned automatically by minimizing an error function between the output class labels and target class labels of the training patterns through gradient descent. The experimental results show that the proposed classifier is valid to be applied to processing training patterns with imprecise class labels. Compared with the k-NN classifier based on the TBM without the parameter optimization, the proposed classifier appears to have lower classification error rates and higher robustness.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2008年第z1期239-243,共5页
Journal of Computer Research and Development
基金
国家自然科学基金项目(60663007)
江西师范大学青年成长基金项目(1731)
