摘要
针对虚拟筛选过程中对接打分函数适应度低及已有数据集不平衡特点,本文提出了基于多目标粒子群优化的对接函数综合评价方法,以求解多个目标和约束条件下的分类器集合权重分配,为解决训练集不平衡状态下分类器优化问题提供了一种有效的方法。实验结果表明该算法提高了虚拟筛选过程中对接打分函数的应用效果。
Focused on the insufficient adaptation of different principles based scoring functions in virtual screening, the paper proposed a novel multi-objective PSO optimized weight vector construction algorithm for classifier ensemble under multiple objects and constraints. The classifying performance of binding decoy conformation from the imbalaneed scoring data showed that the application effect of docking functions in virtual screening is promoted by the algorithm.
出处
《微计算机信息》
2009年第3期197-199,共3页
Control & Automation
基金
国家科技基础条件平台项目"生物信息学网络计算应用系统"(2005DKA64001)