KMeans Clustering

#lannister/machinelearning

k-means算法以k为参数,把n个对象分成k个簇,使簇内具有较高的相似度,而簇间的相似度较低。其处理过程如下:
1.随机选择k个点作为初始的聚类中心;
2.对于剩下的点,根据其与聚类中心的距离,将其归入最近的簇
3.对每个簇,计算所有点的均值作为新的聚类中心
4.重复2、3直到聚类中心不再发生改变

Clustering is to find the relations/connections between data without labels.
K-means is one of the most widely used algorithms.

K-means for non-separated clusters(T-shirt sizing)

• Find closest centroids

  1 2 3 4 5 6 7 8 9 10 11

  K = size(centroids, 1); distance = zeros(K, 1); % to store and return the min distance idx = zeros(size(X,1), 1); for i = 1:size(X, 1) for k = 1:K distance(k) = sqrt(sum((X(i,:)-centroids(k,:)).^2)); end [mini, index] = min(distance); idx(i) = index; end

• Compute Means

  1 2 3 4 5 6 7 8

  [m n] = size(X); centroids = zeros(K, n); for k=1:K log 大专栏   KMeans Clusteringic = idx==k; centroids(k,:) = 1/sum(logic)*sum(X.*logic); % sum(logic) is the number of examples assigned to kth centroid end

• Randomly initialize cluster centroids

  1 2 3 4

  centroids = zeros(K, size(X, 2)); randidx = randperm(size(X, 1)); % Take the first K examples as centroids centroids = X(randidx(1:K), :);

• K-Means Clustering on Pixels

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

  % Run K-Means for i=1:max_iters % Output progress fprintf('K-Means iteration %d/%d...n', i, max_iters); if exist('OCTAVE_VERSION') fflush(stdout); end % For each example in X, assign it to the closest centroid idx = findClosestCentroids(X, centroids); % Given the memberships, compute new centroids centroids = computeCentroids(X, idx, K); end