spectral_clustering_cogs118b-master

  1. 按照式计算相似性矩阵 2. 排除自身的相似度 3. 按照式计算归一化矩阵 4. 按照式计算归一化拉普拉斯图矩阵L 5. 计算L的特征向量，将前个特征值最大的向量按列放置成一个矩阵，即依次为前个特征值最大的特征向量 6. 归一化X形成矩阵Y 7. 对矩阵Y按每行为一个数据点，进行k-means聚类，第行所属的类就是原来所属的类。(1. similarity matrix based on formula computation 2. rule out your own similarity. 3. normalized matrix based on formula computation 4. normalized formula Laplasse matrix L according to formula. 5. Calculate the eigenvectors of L. Place the vectors with the largest eigenvalues in columns into a matrix, that is, the eigenvectors with the largest eigenvalues in turn. 6. normalized X formation matrix Y 7. K-means clustering is applied to matrix Y according to one data point per behavior. The first row belongs to the original class.)