摘要
人工免疫系统是受人体免疫系统启发的一种智能算法,负选择算法作为人工免疫系统的核心算法之一,在各领域被广泛研究和应用。从两方面对负选择算法进行了改进,首先对记忆细胞数量对识别准确率的影响进行了研究,提出一种反馈学习的思想来进行记忆细胞数量的优化,实现提高分类过程中的识别准确率。然后为了解决传统负选择算法存在检测器覆盖空间存在交集、整体覆盖空间较低的问题,提出通过记忆细胞识别半径的自动调整,减少检测器数量,提高整体覆盖空间的方法,这种方法避免了"交叉识别(overlap)"和"识别洞(hole)"现象的出现。最后,实验结果表明算法在解决文本分类问题是有效可行的,其在路透社文本分类数据集上分类准确率达到了93.89%。
Artificial Immune System （AIS） is a type of intelligent algorithm inspired by the principles and processes of the human immune system. As one of the most important algorithms of AIS, negative selection algorithm has been widely used. In order to improve the accuracy of the algorithm and achieve the best classification results when using in text classification, the paper improves the algorithm in two ways： Firstly, the paper introduces a feedback learning method to optimize the number of memory cells dynamically after analyze the impact of the number of memory cells on accuracy. Secondly, the negative selection algorithm has two shortcomings when using in text classification ：the detectors＇ covering space exists intersection and the overall covering space is not complete,these two problems are manifested as the phenomenon of ＂hole＂ and ＂overlap＂ respectively. Hence, the paper proposes the automatic adjustment method to overcome these drawbacks. Finally, the paper conducts text classification comparative experiments, the results show that the proposed algorithm is useful for text classification problem, and achieves a precision of 93.89% on Reuters -21 578 dataset.
出处
《智能计算机与应用》
2013年第5期61-65,共5页
Intelligent Computer and Applications
基金
国家自然科学基金(60975077
90924015)
关键词
负选择算法
人工负选择分类
反馈学习
Negative Selection Algorithm
Artificial Negative Selection Classification
Feedback Learning
