Python的Numpy库中的 nonzero函数，及应用在 hardlim硬限幅函数中

Python的Numpy库中的 nonzero函数，及应用在 hardlim硬限幅函数中

在学习 邓捷的《机器学习  算法原理与编程实践》 中 p168页的 激活函数 hardlim的时候， 看不懂。

查询相关的知识点，学习了下。

先看 python代码

#! /usr/bin/python from numpy import * ''' print a [[-4 -3 -2] [-1 0 1] [ 2 3 4]] print dataset.A [[-4 -3 -2] [-1 0 1] [ 2 3 4]] print dataset.A > 0 [[False False False] [False False True] [ True True True]] print nonzero( dataset.A > 0 ) (array([1, 2, 2, 2]), array([2, 0, 1, 2])) print hardlim( a ) [[0 0 0] [1 1 1] [1 1 1]] ''' def hardlim(dataset): dataset[ nonzero( dataset.A > 0 )[0] ] = 1 dataset[ nonzero( dataset.A <=0 )[0] ] = 0 return dataset if __name__ == "__main__": a = arange(9) - 4 #print a #[-4 -3 -2 -1 0 1 2 3 4] a = a.reshape(3, 3) #print a #[[-4 -3 -2] # [-1 0 1] # [ 2 3 4]] a = mat(a) #print a #[[-4 -3 -2] # [-1 0 1] # [ 2 3 4]] print a print hardlim( a )

numpy是Python用来科学计算的一个非常重要的库，numpy主要用来处理一些矩阵对象

numpy.nonzero()函数，简记如下：

官方文档链接如下：http://docs.scipy.org/doc/numpy/reference/generated/numpy.nonzero.html

内容如下：

numpy.nonzero numpy. nonzero ( a ) Returnthe indices of the elements that are non-zero. Returnsa tuple of arrays, one for each dimension of  a ,containing the indices of the non-zero elements in thatdimension. The values in  a arealways tested and returned in row-major, C-style order. Thecorresponding non-zero values can be obtained with: a[nonzero(a)] Togroup the indices by element, rather than dimension, use: transpose(nonzero(a)) Theresult of this is always a 2-D array, with a row for eachnon-zero element. Parameters: a :array_like Inputarray. Returns: tuple_of_arrays :tuple Indicesof elements that are non-zero. Seealso flatnonzero Returnindices that are non-zero in the flattened version of theinput array. ndarray.nonzero Equivalentndarray method. count_nonzero Countsthe number of non-zero elements in the input array.

举些使用的例子

>>> b1 = np.array([True, False, True, False]) >>> np.nonzero(b1) (array([0, 2]),)

可以看到，对于一维的array，输出的是非零值的“下标”。

>>> import numpy as np >>> a = np.array([[1,2,3],[4,5,6],[7,8,9]]) >>> a > 3 array([[False, False, False], [ True, True, True], [ True, True, True]], dtype=bool) >>> np.nonzero(a > 3) (array([1, 1, 1, 2, 2, 2]), array([0, 1, 2, 0, 1, 2])) >>> np.transpose( np.nonzero(a > 3) ) array([[1, 0], 这就是非零元素的坐标 [1, 1], [1, 2], [2, 0], [2, 1], [2, 2]])

还可以参考：http://www.cnblogs.com/itdyb/p/5766707.html，http://www.th7.cn/Program/Python/201501/351495.shtml