前言

所謂二項(xiàng)分布就是只會(huì)產(chǎn)生兩種結(jié)果的概率

1.概念

下面是一個(gè)二項(xiàng)分布的的theano實(shí)現(xiàn)

import numpy as np
import theano
import theano.tensor as T
from theano.tensor.nnet import conv
from theano.tensor.nnet import softmax
from theano.tensor import shared_randomstreamsx=T.matrix('x')
y =T.fscalars('y')theano.config.floatX='float32'# def dropout_layer(x,y):
#     srng=shared_randomstreams.RandomStreams(np.random.RandomState(0).randint(999))
#     mask=srng.binomial(n=1,p=1-y,size=x.shape)
#     return x*T.cast(mask,theano.config.floatX)a=np.random.randn(2,3)srng=shared_randomstreams.RandomStreams(np.random.RandomState(0).randint(999))
mask=srng.binomial(n=1,p=1-y,size=x.shape)
d=x*T.cast(mask,theano.config.floatX)fun=theano.function([x,y],d)print(fun(a,0.6))

結(jié)果：
可以看到要么是0，要么是1，當(dāng)然是近似