I want to write a sigmoid function:
我想写一个sigmoid函数:
def fn(w, x):
return 1.0 / (np.expm1(-w.dot(x))+0.0)
Because -w.dot(x) is a sparse matrix, I used np.expm1() instead of np.exp(), but how to divide a float by a csr_matrix? Thanks!
因为-w.dot(x)是稀疏矩阵,我使用np.expm1()而不是np.exp(),但是如何用csr_matrix划分浮点数?谢谢!
1 个解决方案
#1
0
from spicy import sparse
res2 = np.expm1(-w.dot(x))
res1 = sparse.csr_matrix(np.ones(res2.shape()))
return res1/res2
#1
0
from spicy import sparse
res2 = np.expm1(-w.dot(x))
res1 = sparse.csr_matrix(np.ones(res2.shape()))
return res1/res2