Neural network ensembles: evaluation of aggregation algorithms
Neural network ensembles: evaluation of aggregation algorithms
复制标题
DOI：
10.1016/j.artint.2004.09.006
复制
发表时间：
2005-04-01
影响因子：
14.4
通讯作者：
Ceccatto, HA
中科院分区：
文献类型：
--
作者：
Granitto, PM;Verdes, PF;Ceccatto, HA
集成的人工神经网络表现出改进的泛化能力，优于那些单一的网络。然而，为了使聚合有效，各个网络必须尽可能准确和多样化。一个重要的问题是，然后，如何调整聚合成员，以便在这两个冲突的条件之间有一个最佳的妥协。在这里，我们提出了一个广泛的评价集成建设的几种算法，包括新的建议，并将它们与文献中的标准方法进行比较。我们还讨论了一个潜在的问题与顺序聚合算法：非频繁的，但破坏性的选择，通过他们的特别糟糕的合奏成员。我们引入修改后的算法，以科普这个问题，允许个人加权的聚合成员。我们的算法和他们的加权修改是有利的测试对其他方法在文献中，产生一个明智的改善性能的标准统计数据库作为基准。(c)2004 Elsevier B.V.保留所有权利。
Ensembles of artificial neural networks show improved generalization capabilities that outperform those of single networks. However, for aggregation to be effective, the individual networks must be as accurate and diverse as possible. An important problem is, then, how to tune the aggregate members in order to have an optimal compromise between these two conflicting conditions. We present here an extensive evaluation of several algorithms for ensemble construction, including new proposals and comparing them with standard methods in the literature. We also discuss a potential problem with sequential aggregation algorithms: the non-frequent but damaging selection through their heuristics of particularly bad ensemble members. We introduce modified algorithms that cope with this problem by allowing individual weighting of aggregate members. Our algorithms and their weighted modifications are favorably tested against other methods in the literature, producing a sensible improvement in performance on most of the standard statistical databases used as benchmarks. (c) 2004 Elsevier B.V. All rights reserved.