lans

说明：  用MATLAB对一图象分别用不同小波分解，观察高低频部分进行重构，比较重构误差，再进行阈值消噪，比较消噪前后图象。(MATLAB right one image with different wavelet decomposition, high-low-frequency observations for some remodeling, more reconstruction error and then de-noising, compared before and after image denoising.)

输入信号五个基于瞬时信息的特征：零中心归一化瞬时幅度功率谱密度的最大值,零中心归一化瞬时幅度绝对值的标准偏差，零中心非弱信号段瞬时相位非线性分量绝对值标准偏差，零中心非弱信号段瞬时相位非线性分量标准偏差，零中心归一化的非弱信号段瞬时频率绝对值的标准偏差，得到识别出的信号类别，本例为2PSK, 4PSK, GMSK, OQPSK, pi/4DQSK中的一个，还可以看Art网络的扩展性。(input signal instantaneous information on five characteristics : In the center is one of the instantaneous rate of the power spectral density of the maximum, is a center to the absolute value of the instantaneous rate of the standard deviation, Non-zero weak signal instantaneous phase of the nonlinear component absolute standard deviation, Non-zero weak signal instantaneous phase of the nonlinear component standard deviation, In the center is one of the non-weak signal of the absolute value of the instantaneous frequency standard deviation, the identification of signal type, The cases of two PSK, 4PSK, GMSK, OQPSK, pi/4DQSK of one, Art can be viewed network scalability.)

通信原理课程做的大作业，关于正交调制的。(communication theory courses do the work, on the orthogonal modulation.)

用蒙特卡洛法模拟编写的计算粒子的多次散射和单次散射的比较(Comparison of multiple scattering method using Monte Carlo simulation calculations prepared particles and single-scattering)

自相关虽然能够判断清/浊音，并估计浊音的基音周期，但是其运算量很大，用乘法运算所需时间较长。为了避免乘法，采用短时平均幅度差函数（AMDF），验证AMDF与自相关有类似的作用，即平均幅度差函数能够代替自相关函数进行语音分析。(Although autocorrelation can differ from the Voiced/Unvoiced and voiced the pitch period estimate, but the operation is large enough, with the multiplication time was needed. In order to avoid the multiplication using the short time average magnitude difference function (AMDF), verify the AMDF and autocorrelation have a similar effect, that the average magnitude difference function can replace the voice of the autocorrelation function.)

MLPG is a combination of collocation method and EFG

最近在看hl2,发现mdl模型必须从36转到37,哈哈,恰好看到此贴,非常感谢。(recently watching hl2, mdl model must be found from 36 to 37, haha, just to see patch. Thank you very much.)

馬可夫鍊的MATLAB程式撰寫和過程 (Markov chains and processes the MATLAB programming)

说明：  matlab可转债定价程序,可以用来对可转债进行定价(matlab for convertible bond pricing model)

说明：  CNN,DBN算法可以对手写体数字进行识别，准确率高(CNN and DBN algorithm can recognize handwritten numerals with high accuracy)