摘要
LogitBoost是直接优化二项对数损失的Adaboost算法。由于其解趋近于贝叶斯最优解，并且实现简单，成为机器学习中研究的热点。文章首次将LogitBoost的分类算法应用于网络误用检测中，实验结果表明，LogitBoost检测性能优于目前常用的模式匹配算法。
LogitBoost is one of the Adaboost algorithms that directly optimizes the binomial log-likelihood. Because its solution approximates to Bayesian and it is easy to implement, LogitBoost becomes the hot topic in machine learning. In this paper, the LogitBoost classification technique is applied in network misuse detection firstly. The result of experiments shows that LogitBoost has better performance than the other common Pattern Match algorithms.
出处
《电子对抗》
2009年第1期21-24,共4页
Electronic Warfare
