BAdaCost: Multi-class Boosting with Costs
BAdaCost: Multi-class Boosting with Costs
复制标题
DOI：
10.1016/j.patcog.2018.02.022
复制
发表时间：
2018-07
期刊：
影响因子：
--
通讯作者：
Antonio Fernández-Baldera;J. M. Buenaposada;L. Baumela
中科院分区：
文献类型：
--
作者：
Antonio Fernández-Baldera;J. M. Buenaposada;L. Baumela
我们提出了BAdaCost，多类成本敏感的分类算法。它结合了一组成本敏感的多类弱学习器，以获得一个强分类规则的Boosting框架。为了推导出该算法，我们引入了CMEL，一个成本敏感的多类指数损失，概括了各种分类算法，如AdaBoost，SAMME，成本敏感的AdaBoost和PIBoost优化的损失。从而将它们统一在一个共同的理论框架下。在实验中，我们证明了BAdaCost实现了显着的性能增益相比，以前的多类成本敏感的方法。该算法在非对称多类分类中的优势也在实际的多视角人脸和汽车检测问题中进行了评估。
We present BAdaCost, a multi-class cost-sensitive classification algorithm. It combines a set of cost-sensitive multi-class weak learners to obtain a strong classification rule within the Boosting framework. To derive the algorithm we introduce CMEL, aCost-sensitive Multi-class Exponential Lossthat generalizes the losses optimized in various classification algorithms such as AdaBoost, SAMME, Cost-sensitive AdaBoost and PIBoost. Hence unifying them under a common theoretical framework. In the experiments performed we prove that BAdaCost achieves significant gains in performance when compared to previous multi-class cost-sensitive approaches. The advantages of the proposed algorithm in asymmetric multi-class classification are also evaluated in practical multi-view face and car detection problems.