（a）产生两个都具有200个二维向量的数据集和（注意：在生成数据集之前最好使用命令randn(‘seed’,0)初始化高斯随机生成器为0（或任意给定数值），这对结果的可重复性很重要）。向量的前半部分来自均值向量的正态分布，并且协方差矩阵。向量的后半部分来自均值向量的正态分布，并且协方差矩阵。其中是一个2*2的单位矩阵。
（b）在上述数据集上运用感知器算法，并且使用不同的初始向量初始化参数向量。
（c）测试每一次算法在和上的性能。
（d）画出数据集和，以及分类面。((a) Generate the sum of two datasets with 200 two-dimensional vectors (Note: before generating the dataset, it is better to initialize the Gaussian random generator to 0 (or any given value) with the command randn ("seed", 0), which is important for the repeatability of the results). The first half of the vector comes from the normal distribution of the mean vector and the covariance matrix. The second half of the vector comes from the normal distribution of the mean vector and the covariance matrix. Where is a 2 * 2 identity matrix.
(b) The perceptron algorithm is used on the above data set, and different initial vectors are used to initialize the parameter vector.
(c) Test the performance of each algorithm on and.
(d) Draw the data set and, as well as the classification surface.)