LinearOpticalElementsGUI

Simulation of m/m/1/k type queuing process

wiener滤波器的实现，主要是学习曲线的绘制和误差的计算(wiener filter implementation, mainly drawing and error learning curve calculation)

对符号LMS算法进行性能仿真，并将其和LMS算法的 误差性能和收敛性能比较(The symbol LMS algorithm performance simulation, and error performance and LMS algorithm and convergence performance comparison)

功率谱分析，经过修改的，yuler_walke、welch、relax、相关图、周期图等(Power spectrum analysis, modified, yuler_walke, welch, relax, related maps, cycle map)

改进的分段线性表示方法，含有四种基本算法，自顶向下，自底向上，滑窗，以及滑窗式自底向上算法。是流行的降维、分类算法。改进之后，所得出的线段是连续的，解决了离散线段带来的不便。(Rewised PLR algorithms inclusing four basic methods: top down, bottom up, sliding window and sliding window bottom up. PLR is a popular algorithm for classification and demension reduction. )

principal component analysis(PCA)主成分分析，Matlab实例代码(principal component analysis(PCA))

Given a matrix with m rows and n cols (m points in R^n), use resampling and the Kolmogorov Smirnov test to score [0,1] all points (as potential outliers) in linear time. This is an original algorithm that can be used for anomaly detection and general signal processing.

tuning of pid controller for a higher order system by different tuing algorithms

说明：  相关向量机（Relevance Vector Machine，简称RVM）是Micnacl E.Tipping于2000年提出的一种与SVM（Support Vector Machine）类似的稀疏概率模型，是一种新的监督学习方法。 它的训练是在贝叶斯框架下进行的，在先验参数的结构下基于主动相关决策理论（automatic relevance determination，简称ARD）来移除不相关的点，从而获得稀疏化的模型。在样本数据的迭代学习过程中，大部分参数的后验分布趋于零，与预测值无关，那些非零参数对应的点被称作相关向量（Relevance Vectors），体现了数据中最核心的特征。同支持向量机相比，相关向量机最大的优点就是极大地减少了核函数的计算量，并且也克服了所选核函数必须满足Mercer条件的缺点。(Relevance Vector Machine (RVM) is a sparse probability model similar to SVM (Support Vector Machine) proposed by Micnacl E. Tipping in 2000. It is a new supervised learning method. Its training is carried out under the Bayesian framework. Under the structure of prior parameters, it is based on Automatic Relevance Determination (ARD) to remove the irrelevant points, so as to obtain the sparse model. In the iterative learning process of sample data, the posterior distribution of most parameters tends to zero, which is independent of the predicted value. The points corresponding to non-zero parameters are called Relevance Vectors, which represent the most core features of the data. Compared with support vector machine, the biggest advantage of correlation vector machine is that it greatly reduces the computation amount of kernel function, and also overcomes the shortcoming that the selected kernel function must meet Mercer's condition.)

说明：  采用白噪声作为输入信号，用递推最小二乘法来进行系统辨识(Using white noise as input signal, using recursive least squares method for system identification)