tensorflow 笔记

tf.one_hot(matrix,depth...)

used to create a matrix converted from original input (classification) using one-hot code.

depth: used to control the output length. e.g.

        input[0,2]

        depth: 3 => [ [1,0,0],[0,0,1]]

        depth: 4 => [[1,0,0,,0],[0,0,1,0]]

tf.boolean_mask(a,b):

will make a matrix containing only the values with the same index in matrix B as "True" or "1" element.