|
|
EDA365欢迎您登录！
您需要 登录 才可以下载或查看，没有帐号？注册
x
# Z% L3 V( }, N% M
通用MATLAB代码+ c2 q0 H% i( N# m4 z
主函数main.m
) T5 V8 v. ?- Y* }' B3 ~% [: }/ g" m4 ~9 c4 q" z5 \- C/ o
- clc
- u1 = rand(4,1);
- u2 = [2;32;6666;5];
- wind_size = size(u1,1);
- mi = calmi(u1, u2, wind_size);" Y/ Q: H7 d8 n8 [5 E3 @# _
% D% G: D* S4 a
; X% O' o" f% ~) dcalmi.m) r0 k: J) Q+ `1 u
, l) I+ u. ~5 B3 u+ R) G- %计算两列向量之间的互信息
- %u1：输入计算的向量1
- %u2：输入计算的向量2
- %wind_size：向量的长度
- function mi = calmi(u1, u2, wind_size)
- x = [u1, u2];
- n = wind_size;
- [xrow, xcol] = size(x);
- bin = zeros(xrow,xcol);
- pmf = zeros(n, 2);
- for i = 1:2
- minx = min(x(:,i));
- maxx = max(x(:,i));
- binwidth = (maxx - minx) / n;
- edges = minx + binwidth*(0:n);
- histcEdges = [-Inf edges(2:end-1) Inf];
- [occur,bin(:,i)] = histc(x(:,i),histcEdges,1); %通过直方图方式计算单个向量的直方图分布
- pmf(:,i) = occur(1:n)./xrow;
- end
- %计算u1和u2的联合概率密度
- jointOccur = accumarray(bin,1,[n,n]); %（xi，yi）两个数据同时落入n*n等分方格中的数量即为联合概率密度
- jointPmf = jointOccur./xrow;
- Hx = -(pmf(:,1))'*log2(pmf(:,1)+eps);
- Hy = -(pmf(:,2))'*log2(pmf(:,2)+eps);
- Hxy = -(jointPmf(：）)'*log2(jointPmf(：）+eps);
- MI = Hx+Hy-Hxy;
- mi = MI/sqrt(Hx*Hy);5 K) I' ]# [1 _# {/ }3 E. w/ Q0 z5 {
3 n1 G8 `' y k+ w2 T |
|
/1 
发表于 2020-11-11 14:52
收藏
淘帖
支持！
反对！