Abstract: A hybrid algorithm based on quantum-behaved particle swarm optimization algorithm and semi-supervised fuzzy kernel clustering algorithm is proposed.It overcomes the drawbacks of fuzzy clustering methods which are sensitive to the initial cluster centers and easily trapped into local minima.The few labeled data can generate correct model with supervised clustering，and then the model aids to guide lots of unlabeled data to clustering，enlarges the numbles of labeled data.Those data that still can’t be labeled are clustered by the fuzzy kernel method based on quantum-behaved particle swarm optimization，and determine marker types.KDD CUP99 data set is implemented to evaluate the proposed algorithm.Compared to other algorithms，the results show the outstanding performance of the proposed algorithm.

Key words: intrusion detection, Quantum-behaved Particle Swarm Optimization（QPSO）, semi-supervised clustering, kernel function

摘要： 提出了一种将量子粒子群优化算法和半监督模糊核聚类算法相结合的混合算法，用以解决入侵检测算法中模糊聚类算法对初始值敏感，容易陷入局部最优的问题。该算法对少量标记数据进行监督聚类得到正确模型，运用这个模型指导大量未标记数据进行聚类，扩充标记数据集合，对仍没有确定标记的数据利用量子粒子群优化的模糊核聚类算法进行聚类，确定其标记类型。通过KDD CUP99实验数据的仿真，实验结果表明，该算法在入侵检测中能获得理想的检测率和误检率。

关键词: 入侵检测, 量子粒子群优化, 半监督聚类, 核函数