assign_2_a

steps - 1. run DigitRecognizer.m 2. mode to run a. Test Mode input pad will open with axes having a red dot. left-click mouse and drag(dont unclick) mouse to draw a numeric digit. if you unclick mouse it won t draw anything. if you want to continue drawing, click again on any red place on axes and start dragging. once done, click submit. it will give you result. b. Training Mode input pad will open with axes having a red dot. draw a number and click submit. when promted, enter the value you drawn. if you wish to continue training, click YES and repeat. if you click No, a feedforward neural network will be trained and saved, you can test it.

实现图像的DCT高通、低通滤波器并与傅里叶变换做比较.有图可见，经DCT低通滤波后，图像变得很模糊。因为DCT是FFT取其实部，能量减少，因此图像变得很模糊，没有经FFT滤波后的效果好。 经DCT高通滤波后的图像： 经DCT高通滤波后，图像的低频部分被滤除掉，图像变黑，只有边缘的痕迹可见。 (The DCT of image high-pass, low pass filter and compared with the Fourier transform.)

matlab软件中有关elman神经网络预测的程序(matlab software procedures elman neural network prediction)

自动保存文件，去开关压制起伏，并显示电离图(auto save)

Matlab粒子群工具箱，实现优化计算中的粒子群算法功能(The Matlab Particle Swarm toolbox, particle swarm optimization calculation algorithm function)

在awgn信道下的的8psk调制，绘出传输和接受的星座图(In awgn channel under the 8psk modulation plotted transmit and receive constellation)

提供了外文文献具体解释了HHT得缺陷之一，模态混叠现象。 通过matlab的仿真具体解释了HHT用于信号分析时出现模态混叠的临界点。(english papers to explain one of the limitations of HHT-MODE MIXING simulation base on MATLAB provide graph to improve that limitation)

Solving SVM through SMO

Gaussian Mixture Utilities

Pe rceptron, con ve rgence, and gene rali zati on Recall tha t we ar e dealing with linear classifiers throug h origin, i.e.