摘要： 以现有组合分类器修剪方法为基础，从增大搜索空间的角度出发，提出一种基于束状搜索的组合分类器修剪方法，在每一步增加或删除一个基分类器时都保存最优的前k个组合。该方法既保持了爬山搜索算法的高效剪枝特性，又能有效减小其过快收敛到局部最优解的可能性，使修剪得到的组合基分类器更接近于全局最优。与传统组合分类器修剪方法的对比结果表明，该方法修剪所得的组合分类器具有更高的分类准确率，并且组合规模也有所降低。

关键词: 组合分类, 组合修剪, 束状搜索, Bagging方法

Abstract: Based on existing ensemble pruning methods, this paper presents a beam search-based combination pruning method from the view of expanding the search. The method adopts beam search method, and saves the first k optimal combinations while adding or removing a search-based ensemble in each step. That not only remains the characteristics of efficiently pruning of the original combination pruning methods in greedy way, but also reduces the risk that fast convergence leads to local optimum easily, makes combination classification after pruned closer to the global optimum. Comparison results show that this method has a higher accuracy of classification, and it is smaller than the original ensemble pruning methods on most datasets.

Key words: combination classification, combination pruning, beam search, Bagging method

中图分类号: 

• TP39