In chapter 3 , by minimizing a quadratic model restricting next search direction to a subset of span { - g _ ( k ) , d _ ( k - 1 ) } , we propose other formula of parameter ~ ( * ) k convergences of the new method and its hybrid formula are proved under monotone line search . their numerical results show that they are efficient 在第三章中，通过最小化一个限制下次搜索方向在span { - g _ k ， d _ ( k - 1 ) }子集上的二次模型，我们提出了_ k ~ *取新形式或其杂交形式的共轭梯度法是有效的。
Otherwise , the trust region radius is reduced and a new trial point is selected . it is possible that the region subproblem need to be resolved many times before obtaining an acceptable step , and hence the total computation for completing one iteration might be expensive . this article combines approximate trust region path and nonmonotonic backtracking strategy to solve nonlinear optimization subject to linear inequality constraints , that is , we use approximate trust region path to get the research direction by minimuming quadratical model in the trust region by employing 本文在求解线性不等式约束优化问题时，将近似信赖域路径与非单调信赖域方法相结合，即在信赖域半径内沿近似信赖域路径得到一极小化二次模型的搜索方向后采用回代法避免重复求解信赖域子问题，在此算法中当搜索方向不被接受时，就用非单调线搜索技术得到接受步长，定义新的迭代点。
It has the important influence on detection rate , false positive rate and ability of real time responds . so this thesis labors over kdd cup 99 intrusion detection data set and uses some rules to reduce the dimensions of feature sample space in the derection of sbg . thus the phenomena of dimension exploding can be avoided 特征提取是入侵检测的关键,对检测率、误报率和实时性有着重要影响,因此本文在针对kddcup99入侵检测数据进行了大量分析,并沿sbg的搜索方向,使用一定的规则集来达到特征样本空间的维数的削减的目的,从而避免了“维数爆炸”的现象。
It guides the search direction according to probabilistic transition regularity . it utilizes the information acquired and accumulated in evolution process to organize search by itself and adaptively control the search direction . it needs little knowledge about search space or other auxiliary information , which fits the data mining task 算法根据概率的变迁规则来指导搜索方向，利用演化过程中获得和积累的有关搜索空间的信息自行组织搜索，并自适应地控制搜索过程，基本上不用搜索空间的知识或其它的辅助信息，对问题本身没有过多的要求，适于数据挖掘的任务环境。
Optimization can be conducted by use of genetic algorithms . this optimization method need n ' t the differential coefficient and other assistant knowledge , but of the object function and the adaptive evaluation function . it ' s powerful to optimize the nonlinear system difficult to set up the mathematical model 遗传算法在优化过程中不需要求导或其它辅助知识，只需要影响搜索方向的目标函数和相应的适应度函数，对难于建立数学模型的非线形系统有较强的优化能力。
At present , constrained optimization methods may be classified to two classes , one is search method which firstly asserts whether the current point is an optimal point , if the point is not , then we must choose search directions , and along the search direction , find the next iterative point which make the objective function or merit function to decrease 目前的约束优化算法可以分成两大类，一类是搜索算法，这种算法首先判断当前点是否为最优点，若非最优点则要确定搜索方向，然后沿此方向确定一个使目标函数或评价函数下降的点。这种算法一般为下降算法，如可行方向法、约束变尺度法等。
It has been deduced that the search direction remains the same but that the step length decreases in optimizing the improved cost function with the conjugate gradient algorithm from the gradient of the cost function that the fir filter apporaches the contrary point spread function ( psf ) more , and that the estimated image is closer to the original one 从代价函数的梯度入手推导出共轭梯度算法，其搜索方向保持不变而搜索步长变小，使滤波系数更加退近于点扩展函数逆算子，从而使估计图像与原始图像更加接近。
Chapter 1 , making use of the technique of pivoting operation , the technique of sequential systems of linear equations of studying constrained optimization problems and the idea of strongly subfeasible directions method and , several new algorithms for systems of nonlinear inequalities are presented 第一章，借助于转轴运算技术，序列线性方程组技术以及强次可行方向法思想，建立了非线性不等式组的序列线性方程组算法，算法的搜索方向由有唯一解的线性方程组的解及其校正产生。
