中文摘要：

已有的基于蚁群优化算法的特征选择方法是从随机点出发，寻找最优的特征组合。讨论和分析了粗糙集理论中的特征核思想，结合蚁群优化算法的全局寻优特点，以特征重要度作为启发式搜索信息，提出从特征核出发基于粗糙集理论与蚁群优化的特征选择算法，简化蚁群完全图搜索的规模。在标准UCI数据集上进行测试，实验验证了新算法对于特征选择的有效性。

英文摘要：

Many existing ACO-based feature selection algorithms start from a random dot,which aim at finding the optimal fea- tures. This thesis analyzed the feature core method of rough sets and the global optimization ability of ACO, proposed a new rough set approach to feature selection based on ACO, which adopted feature significance as heuristic information. The approach started from the feature core, which changed the complete graph to a smaller one. To verify the efficiency of algorithm, carried out experiments on some standard UCI datasets. The results demonstrate that the proposed algorithm can provide efficient solution to find a minimal subset of the features.