Week4_1Neural Networks Representation

Week4_1Neural Networks Representation

第 1 题

Which of the following statements are true? Check all that apply.

  • The activation values of the hidden units in a neural network, with the sigmoid activation function applied at every layer, are always in the range (0, 1).
  • Any logical function over binary-valued (0 or 1) inputs x1 x 1 and x2 x 2 can be (approximately) represented using some neural network.
  • A two layer (one input layer, one output layer; no hidden layer) neural network can represent the XOR function.
  • Suppose you have a multi-class classification problem with three classes, trained with a 3 layer network. Let a(3)1=(hΘ(x))1 be the activation of the first output unit, and similarly a(3)2=(hΘ(x))2 a 2 ( 3 ) = ( h Θ ( x ) ) 2 and a(3)3=(hΘ(x))3 a 3 ( 3 ) = ( h Θ ( x ) ) 3 . Then for any input x, it must be the case that a(3)1+a(3)2+a(3)3=1 a 1 ( 3 ) + a 2 ( 3 ) + a 3 ( 3 ) = 1 .

*     答案: 1 2 *
* 选项1: sigmoid激励层的取值范围都是(0-1). 正确 **
* 选项2: logical function有四种操作:AND OR XOR NOT. 其中两层的Network可以表示出AND OR NOT这三类,三层Network可以表示出XOR. 正确 **
* 选项3: 要实现xor操作须要有一个hidden layer. 不正确 **
* 选项4: 计算三个分类中每一个分类的期望,最理想的情况才会出现类似1 0 0 0这种. 不正确 **


第 2 题

Consider the following neural network which takes two binary-valued inputs x1,x2{0,1} x 1 , x 2 ∈ { 0 , 1 } and outputs hΘ(x) h Θ ( x ) . Which of the following logical functions does it (approximately) compute?
4_1

  • OR
  • AND
  • NAND (meaning “NOT AND”)
  • XOR (exclusive OR)
    *     答案: 1 *
    * hθ=g(20+30x1+30x2) h θ = g ( − 20 + 30 x 1 + 30 x 2 ) *
x1 x 1 x2 x 2 hθ h θ
000
101
011
111

* 只要有一个为1,就为True, 这不就是OR嘛 *


第 3 题

Consider the neural network given below. Which of the following equations correctly computes the activation a(3)1 a 1 ( 3 ) ? Note: g(z) g ( z ) is the sigmoid activation function.
4_1_2
* a(3)1=g(Θ(2)1,0a(2)0+Θ(2)1,1a(2)1+Θ(2)1,2a(2)2) a 1 ( 3 ) = g ( Θ 1 , 0 ( 2 ) a 0 ( 2 ) + Θ 1 , 1 ( 2 ) a 1 ( 2 ) + Θ 1 , 2 ( 2 ) a 2 ( 2 ) )
* a(3)1=g(Θ(1)1,0a(1)0+Θ(1)1,1a(1)1+Θ(1)1,2a(1)2) a 1 ( 3 ) = g ( Θ 1 , 0 ( 1 ) a 0 ( 1 ) + Θ 1 , 1 ( 1 ) a 1 ( 1 ) + Θ 1 , 2 ( 1 ) a 2 ( 1 ) )
* a(3)1=g(Θ(1)1,0a(2)0+Θ(1)1,1a(2)1+Θ(1)1,2a(2)2) a 1 ( 3 ) = g ( Θ 1 , 0 ( 1 ) a 0 ( 2 ) + Θ 1 , 1 ( 1 ) a 1 ( 2 ) + Θ 1 , 2 ( 1 ) a 2 ( 2 ) )
* The activation a(3)1 a 1 ( 3 ) is not present in this network.

*     答案: 1 *
* a(3)1 a 1 ( 3 ) 是第3层的第1个激活单元=第2层的 Θ Θ 乘以 第2层的a *
* 第2层的 Θ Θ 就是 Θ2 Θ 2 *
* 第2层的 a a 就是a2 *


第 4 题

You have the following neural network:
4_1_3
You’d like to compute the activations of the hidden layer a(2)R3 a ( 2 ) ∈ R 3 .
One way to do so is the following Octave code:
4_4_4
You want to have a vectorized implementation of this (i.e., one that does not use for loops). Which of the following implementations correctly compute a(2) a ( 2 ) ? Check all that apply.

  • z = Theta1 * x; a2 = sigmoid (z);
  • a2 = sigmoid (x * Theta1);
  • a2 = sigmoid (Theta2 * x);
  • z = sigmoid(x); a2 = sigmoid (Theta1 * z);

*     答案: 1 *
* 选项2: sigmoid是最后一步, 同时是先 Theta1*x, 再sigmoid(z) **


第 5 题

You are using the neural network pictured below and have learned the parameters Θ(1)=[1111.72.43.2] Θ ( 1 ) = [ 1 1 2.4 1 1.7 3.2 ] (used to compute a(2)) and Θ(2)=[10.31.2] Θ ( 2 ) = [ 1 0.3 − 1.2 ] (used to compute a(3) a ( 3 ) as a function of a(2) a ( 2 ) ).
Suppose you swap the parameters for the first hidden layer between its two units so Θ(1)=[111.713.22.4] Θ ( 1 ) = [ 1 1.7 3.2 1 1 2.4 ] and also swap the output layer so Θ(2)=[11.20.3] Θ ( 2 ) = [ 1 − 1.2 0.3 ] . How will this change the value of the output hΘ(x) h Θ ( x ) ?
4_1_5

  • It will stay the same.
  • It will increase.
  • It will decrease
  • Insufficient information to tell: it may increase or decrease.

*     答案: 1 *
**     当交换 Θ(1) Θ ( 1 ) 中的两行时,实际上就是交换了 a(2)2 a 2 ( 2 ) a(2)3 a 3 ( 2 ) ,
原先是 a(2)10.3 a 1 ( 2 ) ∗ 0.3 + a(2)2(1.2) a 2 ( 2 ) ∗ ( − 1.2 ) 现在变为$a_2^{(2)}(-1.2)+a_1^{(2)}*0.3$$ 没什么区别
*

  • 5
    点赞
  • 5
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值