【Machine Learning】【Andrew Ng】- 编程题(Week 8)

1、Find Closest Centroids

m = size(X,1);
for i = 1:m
   d_min = 10000;
   for j = 1:K
      d_temp = sum((X(i,:)- centroids(j,:)).^2);
      if d_temp<d_min
         d_min = d_temp;
         idx(i) = j;
      end
   end
end

2、Compute Centroid Means

for i = 1:K
    s = 0;
    cnt = 0;
    for j = 1:m
       if idx(j)==i
           s = s + X(j,:);
           cnt = cnt +1;
       end
    end
    centroids(i,:) = s./cnt;
end

这里补充了一个用K均值算法进行图像压缩的程序例子。
3、PCA

Covariance = 1/m*X'*X;
[U,S,D] = svd(Covariance);

4、Project Data

Z = X*U(:,1:K);

X为m×n的矩阵，U为n×n的矩阵，Z为m×K的矩阵。
5、Recover Data

X_rec = Z *U(:,1:K)';

Z为m×K的矩阵，U为n×n的矩阵，X_rec为m×n的矩阵。