python-2.7 - sklearn : semi-supervised learning - LabelSpreadingModel memory error

Question

我正在使用sklearn LabelSpreadingModel，如下所示:

label_spreading_model = LabelSpreading()
model_s = label_spreading_model.fit(my_inputs, labels)

但是我遇到了以下错误:

   MemoryErrorTraceback (most recent call last)
    <ipython-input-17-73adbf1fc908> in <module>()
         11 
         12 label_spreading_model = LabelSpreading()
    ---> 13 model_s = label_spreading_model.fit(my_inputs, labels)

    /usr/local/lib/python2.7/dist-packages/sklearn/semi_supervised/label_propagation.pyc in fit(self, X, y)
        224 
        225         # actual graph construction (implementations should override this)
    --> 226         graph_matrix = self._build_graph()
        227 
        228         # label construction

    /usr/local/lib/python2.7/dist-packages/sklearn/semi_supervised/label_propagation.pyc in _build_graph(self)
        455         affinity_matrix = self._get_kernel(self.X_)
        456         laplacian = graph_laplacian(affinity_matrix, normed=True)
    --> 457         laplacian = -laplacian
        458         if sparse.isspmatrix(laplacian):
        459             diag_mask = (laplacian.row == laplacian.col)

    MemoryError: 

我的输入矩阵的拉普拉斯算子似乎有问题。是否有任何我可以配置的参数或任何可以避免此错误的更改？谢谢!

Answer 1

很明显:您的电脑内存不足。

由于您没有设置任何参数，因此默认使用rbf-kernel ( proof )。

一些摘录自scikit-learn's docs :

The RBF kernel will produce a fully connected graph which is represented in
memory by a dense matrix. This matrix may be very large and combined with the 
cost of performing a full matrix multiplication calculation for each iteration
of the algorithm can lead to prohibitively long running times

也许以下内容(上面文档中的下一句话)会有所帮助:

On the other hand, the KNN kernel will produce a much more memory-friendly 
sparse matrix which can drastically reduce running times.

但我不知道你的数据、电脑配置等。并且只能猜测...