卷积神经网络第一周作业 Convolution+model+-+Application+-+v1

最新推荐文章于 2021-07-30 16:21:59 发布

haoji007

最新推荐文章于 2021-07-30 16:21:59 发布

阅读量376

点赞数

分类专栏：【TensorFlow相关】【深度学习--应用实例】

原文链接：https://blog.csdn.net/Solo95/article/details/85070001

版权

【深度学习--应用实例】同时被 2 个专栏收录

83 篇文章 11 订阅

订阅专栏

【TensorFlow相关】

34 篇文章 0 订阅

订阅专栏

来自吴恩达深度学习系列视频，卷积神经网络第一周作业 Convolution+model±+Application±+v1
如果英文对你来说有困难，请参照：【中文】【吴恩达课后编程作业】Course 4 - 卷积神经网络 - 第一周作业 - 搭建卷积神经网络模型以及应用（1&2）参照对代码的解释并不完全正确请注意这点

在本次作业中，你会在forward_propagation那部分遇到代码完全正确，但是结果和作业提供的答案不一致，这是由于tensorflow版本不一致，继续往下计算出的cost也不一样，但是通过注释了模型里的一行代码:
# seed = seed + 1 #Change made by Steve Wang 博主做的改动
最终模型的准确度会高于旧版本。

完整版ipynb文件见博主github:
https://github.com/Hongze-Wang/Deep-Learning-Andrew-Ng/tree/master/homework

Convolutional Neural Networks: Application

Welcome to Course 4’s second assignment! In this notebook, you will:

Implement helper functions that you will use when implementing a TensorFlow model
Implement a fully functioning ConvNet using TensorFlow

After this assignment you will be able to:

Build and train a ConvNet in TensorFlow for a classification problem

We assume here that you are already familiar with TensorFlow. If you are not, please refer the TensorFlow Tutorial of the third week of Course 2 (“Improving deep neural networks”).

1.0 - TensorFlow model

In the previous assignment, you built helper functions using numpy to understand the mechanics behind convolutional neural networks. Most practical applications of deep learning today are built using programming frameworks, which have many built-in functions you can simply call.

As usual, we will start by loading in the packages.

<span style="color:#000000"><code class="language-python"><span style="color:#c678dd">import</span> math
<span style="color:#c678dd">import</span> numpy <span style="color:#c678dd">as</span> np
<span style="color:#c678dd">import</span> h5py
<span style="color:#c678dd">import</span> matplotlib<span style="color:#999999">.</span>pyplot <span style="color:#c678dd">as</span> plt
<span style="color:#c678dd">import</span> scipy
<span style="color:#c678dd">from</span> PIL <span style="color:#c678dd">import</span> Image
<span style="color:#c678dd">from</span> scipy <span style="color:#c678dd">import</span> ndimage
<span style="color:#c678dd">import</span> tensorflow <span style="color:#c678dd">as</span> tf
<span style="color:#c678dd">from</span> tensorflow<span style="color:#999999">.</span>python<span style="color:#999999">.</span>framework <span style="color:#c678dd">import</span> ops
<span style="color:#c678dd">from</span> cnn_utils <span style="color:#c678dd">import</span> <span style="color:#669900">*</span>

<span style="color:#669900">%</span>matplotlib inline
np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>seed<span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">)</span>
</code></span>

Run the next cell to load the “SIGNS” dataset you are going to use.

<span style="color:#000000"><code class="language-python"><span style="color:#5c6370"># Loading the data (signs)</span>
X_train_orig<span style="color:#999999">,</span> Y_train_orig<span style="color:#999999">,</span> X_test_orig<span style="color:#999999">,</span> Y_test_orig<span style="color:#999999">,</span> classes <span style="color:#669900">=</span> load_dataset<span style="color:#999999">(</span><span style="color:#999999">)</span>
</code></span>

As a reminder, the SIGNS dataset is a collection of 6 signs representing numbers from 0 to 5.

在这里插入图片描述

The next cell will show you an example of a labelled image in the dataset. Feel free to change the value of index below and re-run to see different examples.

<span style="color:#000000"><code class="language-python"><span style="color:#5c6370"># Example of a picture</span>
index <span style="color:#669900">=</span> <span style="color:#98c379">6</span>
plt<span style="color:#999999">.</span>imshow<span style="color:#999999">(</span>X_train_orig<span style="color:#999999">[</span>index<span style="color:#999999">]</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"y = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>np<span style="color:#999999">.</span>squeeze<span style="color:#999999">(</span>Y_train_orig<span style="color:#999999">[</span><span style="color:#999999">:</span><span style="color:#999999">,</span> index<span style="color:#999999">]</span><span style="color:#999999">)</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
</code></span>

y = 2
在这里插入图片描述
In Course 2, you had built a fully-connected network for this dataset. But since this is an image dataset, it is more natural to apply a ConvNet to it.

To get started, let’s examine the shapes of your data.

<span style="color:#000000"><code class="language-python"> <span style="color:#669900">"""
def convert_to_one_hot(Y, C):
    Y = np.eye(C)[Y.reshape(-1)].T
    return Y
"""</span>

X_train <span style="color:#669900">=</span> X_train_orig<span style="color:#669900">/</span><span style="color:#98c379">255</span><span style="color:#999999">.</span>
X_test <span style="color:#669900">=</span> X_test_orig<span style="color:#669900">/</span><span style="color:#98c379">255</span><span style="color:#999999">.</span>
Y_train <span style="color:#669900">=</span> convert_to_one_hot<span style="color:#999999">(</span>Y_train_orig<span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span><span style="color:#999999">.</span>T
Y_test <span style="color:#669900">=</span> convert_to_one_hot<span style="color:#999999">(</span>Y_test_orig<span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span><span style="color:#999999">.</span>T
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"number of training examples = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>X_train<span style="color:#999999">.</span>shape<span style="color:#999999">[</span><span style="color:#98c379">0</span><span style="color:#999999">]</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"number of test examples = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>X_test<span style="color:#999999">.</span>shape<span style="color:#999999">[</span><span style="color:#98c379">0</span><span style="color:#999999">]</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"X_train shape: "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>X_train<span style="color:#999999">.</span>shape<span style="color:#999999">)</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"Y_train shape: "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>Y_train<span style="color:#999999">.</span>shape<span style="color:#999999">)</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"X_test shape: "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>X_test<span style="color:#999999">.</span>shape<span style="color:#999999">)</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"Y_test shape: "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>Y_test<span style="color:#999999">.</span>shape<span style="color:#999999">)</span><span style="color:#999999">)</span>
conv_layers <span style="color:#669900">=</span> <span style="color:#999999">{</span><span style="color:#999999">}</span>
</code></span>

<span style="color:#000000"><code class="language-python">number of training examples <span style="color:#669900">=</span> <span style="color:#98c379">1080</span>
number of test examples <span style="color:#669900">=</span> <span style="color:#98c379">120</span>
X_train shape<span style="color:#999999">:</span> <span style="color:#999999">(</span><span style="color:#98c379">1080</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">3</span><span style="color:#999999">)</span>
Y_train shape<span style="color:#999999">:</span> <span style="color:#999999">(</span><span style="color:#98c379">1080</span><span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span>
X_test shape<span style="color:#999999">:</span> <span style="color:#999999">(</span><span style="color:#98c379">120</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">3</span><span style="color:#999999">)</span>
Y_test shape<span style="color:#999999">:</span> <span style="color:#999999">(</span><span style="color:#98c379">120</span><span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span>
</code></span>

1.1 - Create placeholders

TensorFlow requires that you create placeholders for the input data that will be fed into the model when running the session.

Exercise: Implement the function below to create placeholders for the input image X and the output Y. You should not define the number of training examples for the moment. To do so, you could use “None” as the batch size, it will give you the flexibility to choose it later. Hence X should be of dimension [None, n_H0, n_W0, n_C0] and Y should be of dimension [None, n_y]. Hint.

<span style="color:#000000"><code class="language-python"><span style="color:#5c6370"># GRADED FUNCTION: create_placeholders</span>

<span style="color:#c678dd">def</span> <span style="color:#61aeee">create_placeholders</span><span style="color:#999999">(</span>n_H0<span style="color:#999999">,</span> n_W0<span style="color:#999999">,</span> n_C0<span style="color:#999999">,</span> n_y<span style="color:#999999">)</span><span style="color:#999999">:</span>
    <span style="color:#669900">"""
    Creates the placeholders for the tensorflow session.
    
    Arguments:
    n_H0 -- scalar, height of an input image
    n_W0 -- scalar, width of an input image
    n_C0 -- scalar, number of channels of the input
    n_y -- scalar, number of classes
        
    Returns:
    X -- placeholder for the data input, of shape [None, n_H0, n_W0, n_C0] and dtype "float"
    Y -- placeholder for the input labels, of shape [None, n_y] and dtype "float"
    """</span>

    <span style="color:#5c6370">### START CODE HERE ### (≈2 lines)</span>
    X <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>placeholder<span style="color:#999999">(</span>tf<span style="color:#999999">.</span>float32<span style="color:#999999">,</span> <span style="color:#999999">[</span><span style="color:#56b6c2">None</span><span style="color:#999999">,</span> n_H0<span style="color:#999999">,</span> n_W0<span style="color:#999999">,</span> n_C0<span style="color:#999999">]</span><span style="color:#999999">)</span>
    Y <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>placeholder<span style="color:#999999">(</span>tf<span style="color:#999999">.</span>float32<span style="color:#999999">,</span> <span style="color:#999999">[</span><span style="color:#56b6c2">None</span><span style="color:#999999">,</span> n_y<span style="color:#999999">]</span><span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#c678dd">return</span> X<span style="color:#999999">,</span> Y
</code></span>

<span style="color:#000000"><code class="language-python">X<span style="color:#999999">,</span> Y <span style="color:#669900">=</span> create_placeholders<span style="color:#999999">(</span><span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">3</span><span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"X = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>X<span style="color:#999999">)</span><span style="color:#999999">)</span>
<span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"Y = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>Y<span style="color:#999999">)</span><span style="color:#999999">)</span>
</code></span>

<span style="color:#000000"><code>X = Tensor("Placeholder:0", shape=(?, 64, 64, 3), dtype=float32)
Y = Tensor("Placeholder_1:0", shape=(?, 6), dtype=float32)
</code></span>

1.2 - Initialize parameters

You will initialize weights/filters W1W1W1 and W2W2W2 using tf.contrib.layers.xavier_initializer(seed = 0). You don’t need to worry about bias variables as you will soon see that TensorFlow functions take care of the bias. Note also that you will only initialize the weights/filters for the conv2d functions. TensorFlow initializes the layers for the fully connected part automatically. We will talk more about that later in this assignment.

Exercise: Implement initialize_parameters(). The dimensions for each group of filters are provided below. Reminder - to initialize a parameter WWW of shape [1,2,3,4] in Tensorflow, use:

<span style="color:#000000"><code class="language-python">W <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>get_variable<span style="color:#999999">(</span><span style="color:#669900">"W"</span><span style="color:#999999">,</span> <span style="color:#999999">[</span><span style="color:#98c379">1</span><span style="color:#999999">,</span><span style="color:#98c379">2</span><span style="color:#999999">,</span><span style="color:#98c379">3</span><span style="color:#999999">,</span><span style="color:#98c379">4</span><span style="color:#999999">]</span><span style="color:#999999">,</span> initializer <span style="color:#669900">=</span> <span style="color:#999999">.</span><span style="color:#999999">.</span><span style="color:#999999">.</span><span style="color:#999999">)</span>
</code></span>

More Info.

<span style="color:#000000"><code class="language-python"><span style="color:#5c6370"># GRADED FUNCTION: initialize_parameters</span>

<span style="color:#c678dd">def</span> <span style="color:#61aeee">initialize_parameters</span><span style="color:#999999">(</span><span style="color:#999999">)</span><span style="color:#999999">:</span>
    <span style="color:#669900">"""
    Initializes weight parameters to build a neural network with tensorflow. The shapes are:
                        W1 : [4, 4, 3, 8]
                        W2 : [2, 2, 8, 16]
    Returns:
    parameters -- a dictionary of tensors containing W1, W2
    """</span>
    
    tf<span style="color:#999999">.</span>set_random_seed<span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">)</span>                              <span style="color:#5c6370"># so that your "random" numbers match ours</span>
        
    <span style="color:#5c6370">### START CODE HERE ### (approx. 2 lines of code)</span>
    W1 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>get_variable<span style="color:#999999">(</span><span style="color:#669900">"W1"</span><span style="color:#999999">,</span> <span style="color:#999999">[</span><span style="color:#98c379">4</span><span style="color:#999999">,</span> <span style="color:#98c379">4</span><span style="color:#999999">,</span> <span style="color:#98c379">3</span><span style="color:#999999">,</span> <span style="color:#98c379">8</span><span style="color:#999999">]</span><span style="color:#999999">,</span> initializer<span style="color:#669900">=</span>tf<span style="color:#999999">.</span>contrib<span style="color:#999999">.</span>layers<span style="color:#999999">.</span>xavier_initializer<span style="color:#999999">(</span>seed<span style="color:#669900">=</span><span style="color:#98c379">0</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
    W2 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>get_variable<span style="color:#999999">(</span><span style="color:#669900">"W2"</span><span style="color:#999999">,</span> <span style="color:#999999">[</span><span style="color:#98c379">2</span><span style="color:#999999">,</span> <span style="color:#98c379">2</span><span style="color:#999999">,</span> <span style="color:#98c379">8</span><span style="color:#999999">,</span> <span style="color:#98c379">16</span><span style="color:#999999">]</span><span style="color:#999999">,</span> initializer<span style="color:#669900">=</span>tf<span style="color:#999999">.</span>contrib<span style="color:#999999">.</span>layers<span style="color:#999999">.</span>xavier_initializer<span style="color:#999999">(</span>seed<span style="color:#669900">=</span><span style="color:#98c379">0</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>

    parameters <span style="color:#669900">=</span> <span style="color:#999999">{</span><span style="color:#669900">"W1"</span><span style="color:#999999">:</span> W1<span style="color:#999999">,</span>
                  <span style="color:#669900">"W2"</span><span style="color:#999999">:</span> W2<span style="color:#999999">}</span>
    
    <span style="color:#c678dd">return</span> parameters
</code></span>

<span style="color:#000000"><code class="language-python">tf<span style="color:#999999">.</span>reset_default_graph<span style="color:#999999">(</span><span style="color:#999999">)</span>
<span style="color:#c678dd">with</span> tf<span style="color:#999999">.</span>Session<span style="color:#999999">(</span><span style="color:#999999">)</span> <span style="color:#c678dd">as</span> sess_test<span style="color:#999999">:</span>
    parameters <span style="color:#669900">=</span> initialize_parameters<span style="color:#999999">(</span><span style="color:#999999">)</span>
    init <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>global_variables_initializer<span style="color:#999999">(</span><span style="color:#999999">)</span>
    sess_test<span style="color:#999999">.</span>run<span style="color:#999999">(</span>init<span style="color:#999999">)</span>
    <span style="color:#c678dd">print</span><span style="color:#999999">(</span><span style="color:#669900">"W1 = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>parameters<span style="color:#999999">[</span><span style="color:#669900">"W1"</span><span style="color:#999999">]</span><span style="color:#999999">.</span><span style="color:#669900">eval</span><span style="color:#999999">(</span><span style="color:#999999">)</span><span style="color:#999999">[</span><span style="color:#98c379">1</span><span style="color:#999999">,</span><span style="color:#98c379">1</span><span style="color:#999999">,</span><span style="color:#98c379">1</span><span style="color:#999999">]</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
    <span style="color:#c678dd">print</span><span style="color:#999999">(</span><span style="color:#669900">"W2 = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>parameters<span style="color:#999999">[</span><span style="color:#669900">"W2"</span><span style="color:#999999">]</span><span style="color:#999999">.</span><span style="color:#669900">eval</span><span style="color:#999999">(</span><span style="color:#999999">)</span><span style="color:#999999">[</span><span style="color:#98c379">1</span><span style="color:#999999">,</span><span style="color:#98c379">1</span><span style="color:#999999">,</span><span style="color:#98c379">1</span><span style="color:#999999">]</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
</code></span>

<span style="color:#000000"><code>W1 = [ 0.00131723  0.1417614  -0.04434952  0.09197326  0.14984085 -0.03514394
 -0.06847463  0.05245192]
W2 = [-0.08566415  0.17750949  0.11974221  0.16773748 -0.0830943  -0.08058
 -0.00577033 -0.14643836  0.24162132 -0.05857408 -0.19055021  0.1345228
 -0.22779644 -0.1601823  -0.16117483 -0.10286498]
</code></span>

1.2 - Forward propagation

In TensorFlow, there are built-in functions that carry out the convolution steps for you.

tf.nn.conv2d(X,W1, strides = [1,s,s,1], padding = ‘SAME’): given an input XXX and a group of filters W1W1W1, this function convolves W1W1W1's filters on X. The third input ([1,f,f,1]) represents the strides for each dimension of the input (m, n_H_prev, n_W_prev, n_C_prev). You can read the full documentation here
tf.nn.max_pool(A, ksize = [1,f,f,1], strides = [1,s,s,1], padding = ‘SAME’): given an input A, this function uses a window of size (f, f) and strides of size (s, s) to carry out max pooling over each window. You can read the full documentation here
tf.nn.relu(Z1): computes the elementwise ReLU of Z1 (which can be any shape). You can read the full documentation here.
tf.contrib.layers.flatten§: given an input P, this function flattens each example into a 1D vector it while maintaining the batch-size. It returns a flattened tensor with shape [batch_size, k]. You can read the full documentation here.
tf.contrib.layers.fully_connected(F, num_outputs): given a the flattened input F, it returns the output computed using a fully connected layer. You can read the full documentation here.

In the last function above (tf.contrib.layers.fully_connected), the fully connected layer automatically initializes weights in the graph and keeps on training them as you train the model. Hence, you did not need to initialize those weights when initializing the parameters.

Exercise:

Implement the forward_propagation function below to build the following model: CONV2D -> RELU -> MAXPOOL -> CONV2D -> RELU -> MAXPOOL -> FLATTEN -> FULLYCONNECTED. You should use the functions above.

In detail, we will use the following parameters for all the steps:
- Conv2D: stride 1, padding is “SAME”
- ReLU
- Max pool: Use an 8 by 8 filter size and an 8 by 8 stride, padding is “SAME”
- Conv2D: stride 1, padding is “SAME”
- ReLU
- Max pool: Use a 4 by 4 filter size and a 4 by 4 stride, padding is “SAME”
- Flatten the previous output.
- FULLYCONNECTED (FC) layer: Apply a fully connected layer without an non-linear activation function. Do not call the softmax here. This will result in 6 neurons in the output layer, which then get passed later to a softmax. In TensorFlow, the softmax and cost function are lumped together into a single function, which you’ll call in a different function when computing the cost.

<span style="color:#000000"><code class="language-python">
```python
<span style="color:#5c6370"># GRADED FUNCTION: forward_propagation</span>

<span style="color:#c678dd">def</span> <span style="color:#61aeee">forward_propagation</span><span style="color:#999999">(</span>X<span style="color:#999999">,</span> parameters<span style="color:#999999">)</span><span style="color:#999999">:</span>
    <span style="color:#669900">"""
    Implements the forward propagation for the model:
    CONV2D -> RELU -> MAXPOOL -> CONV2D -> RELU -> MAXPOOL -> FLATTEN -> FULLYCONNECTED
    
    Arguments:
    X -- input dataset placeholder, of shape (input size, number of examples)
    parameters -- python dictionary containing your parameters "W1", "W2"
                  the shapes are given in initialize_parameters

    Returns:
    Z3 -- the output of the last LINEAR unit
    """</span>
    
    <span style="color:#5c6370"># Retrieve the parameters from the dictionary "parameters" </span>
    W1 <span style="color:#669900">=</span> parameters<span style="color:#999999">[</span><span style="color:#669900">'W1'</span><span style="color:#999999">]</span>
    W2 <span style="color:#669900">=</span> parameters<span style="color:#999999">[</span><span style="color:#669900">'W2'</span><span style="color:#999999">]</span>
    
    <span style="color:#5c6370">### START CODE HERE ###</span>
    <span style="color:#5c6370"># CONV2D: stride of 1, padding 'SAME'</span>
    Z1 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>conv2d<span style="color:#999999">(</span>X<span style="color:#999999">,</span> W1<span style="color:#999999">,</span> strides<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">,</span> padding<span style="color:#669900">=</span><span style="color:#669900">'SAME'</span><span style="color:#999999">)</span>
    <span style="color:#5c6370"># RELU</span>
    
    A1 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>relu<span style="color:#999999">(</span>Z1<span style="color:#999999">)</span>
    <span style="color:#5c6370"># MAXPOOL: window 8x8, sride 8, padding 'SAME'</span>
    P1 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>max_pool<span style="color:#999999">(</span>A1<span style="color:#999999">,</span> ksize<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">8</span><span style="color:#999999">,</span> <span style="color:#98c379">8</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">,</span> strides<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">8</span><span style="color:#999999">,</span> <span style="color:#98c379">8</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">,</span> padding<span style="color:#669900">=</span><span style="color:#669900">'SAME'</span><span style="color:#999999">)</span>
    <span style="color:#5c6370"># CONV2D: filters W2, stride 1, padding 'SAME'</span>
    Z2 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>conv2d<span style="color:#999999">(</span>P1<span style="color:#999999">,</span> W2<span style="color:#999999">,</span> strides<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">,</span> padding<span style="color:#669900">=</span><span style="color:#669900">'SAME'</span><span style="color:#999999">)</span>
    <span style="color:#5c6370"># RELU</span>
    A2 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>relu<span style="color:#999999">(</span>Z2<span style="color:#999999">)</span>
    <span style="color:#5c6370"># MAXPOOL: window 4x4, stride 4, padding 'SAME'</span>
    P2 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>max_pool<span style="color:#999999">(</span>A2<span style="color:#999999">,</span> ksize<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">4</span><span style="color:#999999">,</span> <span style="color:#98c379">4</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">,</span> strides<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">,</span> <span style="color:#98c379">4</span><span style="color:#999999">,</span> <span style="color:#98c379">4</span><span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">,</span> padding<span style="color:#669900">=</span><span style="color:#669900">'SAME'</span><span style="color:#999999">)</span>
    <span style="color:#5c6370"># FLATTEN</span>
    F <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>contrib<span style="color:#999999">.</span>layers<span style="color:#999999">.</span>flatten<span style="color:#999999">(</span>P2<span style="color:#999999">)</span>
    <span style="color:#5c6370"># FULLY-CONNECTED without non-linear activation function (not not call softmax).</span>
    <span style="color:#5c6370"># 6 neurons in output layer. Hint: one of the arguments should be "activation_fn=None" </span>
    Z3 <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>contrib<span style="color:#999999">.</span>layers<span style="color:#999999">.</span>fully_connected<span style="color:#999999">(</span>F<span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">,</span> activation_fn<span style="color:#669900">=</span><span style="color:#56b6c2">None</span><span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>

    <span style="color:#c678dd">return</span> Z3
</code></span>

<span style="color:#000000"><code class="language-python">tf<span style="color:#999999">.</span>reset_default_graph<span style="color:#999999">(</span><span style="color:#999999">)</span>

<span style="color:#c678dd">with</span> tf<span style="color:#999999">.</span>Session<span style="color:#999999">(</span><span style="color:#999999">)</span> <span style="color:#c678dd">as</span> sess<span style="color:#999999">:</span>
    np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>seed<span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">)</span>
    X<span style="color:#999999">,</span> Y <span style="color:#669900">=</span> create_placeholders<span style="color:#999999">(</span><span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">3</span><span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span>
    parameters <span style="color:#669900">=</span> initialize_parameters<span style="color:#999999">(</span><span style="color:#999999">)</span>
    Z3 <span style="color:#669900">=</span> forward_propagation<span style="color:#999999">(</span>X<span style="color:#999999">,</span> parameters<span style="color:#999999">)</span>
    init <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>global_variables_initializer<span style="color:#999999">(</span><span style="color:#999999">)</span>
    sess<span style="color:#999999">.</span>run<span style="color:#999999">(</span>init<span style="color:#999999">)</span>
    a <span style="color:#669900">=</span> sess<span style="color:#999999">.</span>run<span style="color:#999999">(</span>Z3<span style="color:#999999">,</span> <span style="color:#999999">{</span>X<span style="color:#999999">:</span> np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>randn<span style="color:#999999">(</span><span style="color:#98c379">2</span><span style="color:#999999">,</span><span style="color:#98c379">64</span><span style="color:#999999">,</span><span style="color:#98c379">64</span><span style="color:#999999">,</span><span style="color:#98c379">3</span><span style="color:#999999">)</span><span style="color:#999999">,</span> Y<span style="color:#999999">:</span> np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>randn<span style="color:#999999">(</span><span style="color:#98c379">2</span><span style="color:#999999">,</span><span style="color:#98c379">6</span><span style="color:#999999">)</span><span style="color:#999999">}</span><span style="color:#999999">)</span>
    <span style="color:#c678dd">print</span><span style="color:#999999">(</span><span style="color:#669900">"Z3 = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>a<span style="color:#999999">)</span><span style="color:#999999">)</span>
</code></span>

<span style="color:#000000"><code>Z3 = [[ 1.4416984  -0.24909666  5.450499   -0.2618962  -0.20669907  1.3654671 ]
 [ 1.4070846  -0.02573211  5.08928    -0.48669922 -0.40940708  1.2624859 ]]
</code></span>

在这里插入图片描述
注意从这里开始与预期的输出不一致，是tensorflow的版本不一致造成的，你的结果和我一样表示你的代码并没有问题。

1.3 - Compute cost

Implement the compute cost function below. You might find these two functions helpful:

tf.nn.softmax_cross_entropy_with_logits(logits = Z3, labels = Y): computes the softmax entropy loss. This function both computes the softmax activation function as well as the resulting loss. You can check the full documentation here.
tf.reduce_mean: computes the mean of elements across dimensions of a tensor. Use this to sum the losses over all the examples to get the overall cost. You can check the full documentation here.

** Exercise**: Compute the cost below using the function above.

<span style="color:#000000"><code class="language-python"><span style="color:#5c6370"># GRADED FUNCTION: compute_cost </span>

<span style="color:#c678dd">def</span> <span style="color:#61aeee">compute_cost</span><span style="color:#999999">(</span>Z3<span style="color:#999999">,</span> Y<span style="color:#999999">)</span><span style="color:#999999">:</span>
    <span style="color:#669900">"""
    Computes the cost
    
    Arguments:
    Z3 -- output of forward propagation (output of the last LINEAR unit), of shape (6, number of examples)
    Y -- "true" labels vector placeholder, same shape as Z3
    
    Returns:
    cost - Tensor of the cost function
    """</span>
    
    <span style="color:#5c6370">### START CODE HERE ### (1 line of code)</span>
    cost <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>reduce_mean<span style="color:#999999">(</span>tf<span style="color:#999999">.</span>nn<span style="color:#999999">.</span>softmax_cross_entropy_with_logits_v2<span style="color:#999999">(</span>logits<span style="color:#669900">=</span>Z3<span style="color:#999999">,</span> labels<span style="color:#669900">=</span>Y<span style="color:#999999">)</span><span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#c678dd">return</span> cost
</code></span>

<span style="color:#000000"><code class="language-python">tf<span style="color:#999999">.</span>reset_default_graph<span style="color:#999999">(</span><span style="color:#999999">)</span>

<span style="color:#c678dd">with</span> tf<span style="color:#999999">.</span>Session<span style="color:#999999">(</span><span style="color:#999999">)</span> <span style="color:#c678dd">as</span> sess<span style="color:#999999">:</span>
    np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>seed<span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">)</span>
    X<span style="color:#999999">,</span> Y <span style="color:#669900">=</span> create_placeholders<span style="color:#999999">(</span><span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> <span style="color:#98c379">3</span><span style="color:#999999">,</span> <span style="color:#98c379">6</span><span style="color:#999999">)</span>
    parameters <span style="color:#669900">=</span> initialize_parameters<span style="color:#999999">(</span><span style="color:#999999">)</span>
    Z3 <span style="color:#669900">=</span> forward_propagation<span style="color:#999999">(</span>X<span style="color:#999999">,</span> parameters<span style="color:#999999">)</span>
    cost <span style="color:#669900">=</span> compute_cost<span style="color:#999999">(</span>Z3<span style="color:#999999">,</span> Y<span style="color:#999999">)</span>
    init <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>global_variables_initializer<span style="color:#999999">(</span><span style="color:#999999">)</span>
    sess<span style="color:#999999">.</span>run<span style="color:#999999">(</span>init<span style="color:#999999">)</span>
    a <span style="color:#669900">=</span> sess<span style="color:#999999">.</span>run<span style="color:#999999">(</span>cost<span style="color:#999999">,</span> <span style="color:#999999">{</span>X<span style="color:#999999">:</span> np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>randn<span style="color:#999999">(</span><span style="color:#98c379">4</span><span style="color:#999999">,</span><span style="color:#98c379">64</span><span style="color:#999999">,</span><span style="color:#98c379">64</span><span style="color:#999999">,</span><span style="color:#98c379">3</span><span style="color:#999999">)</span><span style="color:#999999">,</span> Y<span style="color:#999999">:</span> np<span style="color:#999999">.</span>random<span style="color:#999999">.</span>randn<span style="color:#999999">(</span><span style="color:#98c379">4</span><span style="color:#999999">,</span><span style="color:#98c379">6</span><span style="color:#999999">)</span><span style="color:#999999">}</span><span style="color:#999999">)</span>
    <span style="color:#c678dd">print</span><span style="color:#999999">(</span><span style="color:#669900">"cost = "</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>a<span style="color:#999999">)</span><span style="color:#999999">)</span>
</code></span>

<span style="color:#000000"><code>cost = 4.6648693
</code></span>

1.4 Model

Finally you will merge the helper functions you implemented above to build a model. You will train it on the SIGNS dataset.

You have implemented random_mini_batches() in the Optimization programming assignment of course 2. Remember that this function returns a list of mini-batches.

Exercise: Complete the function below.

The model below should:

create placeholders
initialize parameters
forward propagate
compute the cost
create an optimizer

Finally you will create a session and run a for loop for num_epochs, get the mini-batches, and then for each mini-batch you will optimize the function. Hint for initializing the variables

<span style="color:#000000"><code class="language-python"><span style="color:#5c6370"># GRADED FUNCTION: model</span>

<span style="color:#c678dd">def</span> <span style="color:#61aeee">model</span><span style="color:#999999">(</span>X_train<span style="color:#999999">,</span> Y_train<span style="color:#999999">,</span> X_test<span style="color:#999999">,</span> Y_test<span style="color:#999999">,</span> learning_rate <span style="color:#669900">=</span> <span style="color:#98c379">0.009</span><span style="color:#999999">,</span>
          num_epochs <span style="color:#669900">=</span> <span style="color:#98c379">100</span><span style="color:#999999">,</span> minibatch_size <span style="color:#669900">=</span> <span style="color:#98c379">64</span><span style="color:#999999">,</span> print_cost <span style="color:#669900">=</span> <span style="color:#56b6c2">True</span><span style="color:#999999">)</span><span style="color:#999999">:</span>
    <span style="color:#669900">"""
    Implements a three-layer ConvNet in Tensorflow:
    CONV2D -> RELU -> MAXPOOL -> CONV2D -> RELU -> MAXPOOL -> FLATTEN -> FULLYCONNECTED
    
    Arguments:
    X_train -- training set, of shape (None, 64, 64, 3)
    Y_train -- test set, of shape (None, n_y = 6)
    X_test -- training set, of shape (None, 64, 64, 3)
    Y_test -- test set, of shape (None, n_y = 6)
    learning_rate -- learning rate of the optimization
    num_epochs -- number of epochs of the optimization loop
    minibatch_size -- size of a minibatch
    print_cost -- True to print the cost every 100 epochs
    
    Returns:
    train_accuracy -- real number, accuracy on the train set (X_train)
    test_accuracy -- real number, testing accuracy on the test set (X_test)
    parameters -- parameters learnt by the model. They can then be used to predict.
    """</span>
    
    ops<span style="color:#999999">.</span>reset_default_graph<span style="color:#999999">(</span><span style="color:#999999">)</span>                         <span style="color:#5c6370"># to be able to rerun the model without overwriting tf variables</span>
    tf<span style="color:#999999">.</span>set_random_seed<span style="color:#999999">(</span><span style="color:#98c379">1</span><span style="color:#999999">)</span>                             <span style="color:#5c6370"># to keep results consistent (tensorflow seed)</span>
    seed <span style="color:#669900">=</span> <span style="color:#98c379">3</span>                                          <span style="color:#5c6370"># to keep results consistent (numpy seed)</span>
    <span style="color:#999999">(</span>m<span style="color:#999999">,</span> n_H0<span style="color:#999999">,</span> n_W0<span style="color:#999999">,</span> n_C0<span style="color:#999999">)</span> <span style="color:#669900">=</span> X_train<span style="color:#999999">.</span>shape             
    n_y <span style="color:#669900">=</span> Y_train<span style="color:#999999">.</span>shape<span style="color:#999999">[</span><span style="color:#98c379">1</span><span style="color:#999999">]</span>                            
    costs <span style="color:#669900">=</span> <span style="color:#999999">[</span><span style="color:#999999">]</span>                                        <span style="color:#5c6370"># To keep track of the cost</span>
    
    <span style="color:#5c6370"># Create Placeholders of the correct shape</span>
    <span style="color:#5c6370">### START CODE HERE ### (1 line)</span>
    X<span style="color:#999999">,</span> Y <span style="color:#669900">=</span> create_placeholders<span style="color:#999999">(</span>n_H0<span style="color:#999999">,</span> n_W0<span style="color:#999999">,</span> n_C0<span style="color:#999999">,</span> n_y<span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#5c6370"># Initialize parameters</span>
    <span style="color:#5c6370">### START CODE HERE ### (1 line)</span>
    parameters <span style="color:#669900">=</span> initialize_parameters<span style="color:#999999">(</span><span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#5c6370"># Forward propagation: Build the forward propagation in the tensorflow graph</span>
    <span style="color:#5c6370">### START CODE HERE ### (1 line)</span>
    Z3 <span style="color:#669900">=</span> forward_propagation<span style="color:#999999">(</span>X<span style="color:#999999">,</span> parameters<span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#5c6370"># Cost function: Add cost function to tensorflow graph</span>
    <span style="color:#5c6370">### START CODE HERE ### (1 line)</span>
    cost <span style="color:#669900">=</span> compute_cost<span style="color:#999999">(</span>Z3<span style="color:#999999">,</span> Y<span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#5c6370"># Backpropagation: Define the tensorflow optimizer. Use an AdamOptimizer that minimizes the cost.</span>
    <span style="color:#5c6370">### START CODE HERE ### (1 line)</span>
    optimizer <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>train<span style="color:#999999">.</span>AdamOptimizer<span style="color:#999999">(</span>learning_rate<span style="color:#669900">=</span>learning_rate<span style="color:#999999">)</span><span style="color:#999999">.</span>minimize<span style="color:#999999">(</span>cost<span style="color:#999999">)</span>
    <span style="color:#5c6370">### END CODE HERE ###</span>
    
    <span style="color:#5c6370"># Initialize all the variables globally</span>
    init <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>global_variables_initializer<span style="color:#999999">(</span><span style="color:#999999">)</span>
     
    <span style="color:#5c6370"># Start the session to compute the tensorflow graph</span>
    <span style="color:#c678dd">with</span> tf<span style="color:#999999">.</span>Session<span style="color:#999999">(</span><span style="color:#999999">)</span> <span style="color:#c678dd">as</span> sess<span style="color:#999999">:</span>
        
        <span style="color:#5c6370"># Run the initialization</span>
        sess<span style="color:#999999">.</span>run<span style="color:#999999">(</span>init<span style="color:#999999">)</span>
        
        <span style="color:#5c6370"># Do the training loop</span>
        <span style="color:#c678dd">for</span> epoch <span style="color:#c678dd">in</span> <span style="color:#669900">range</span><span style="color:#999999">(</span>num_epochs<span style="color:#999999">)</span><span style="color:#999999">:</span>

            minibatch_cost <span style="color:#669900">=</span> <span style="color:#98c379">0</span><span style="color:#999999">.</span>
            num_minibatches <span style="color:#669900">=</span> <span style="color:#669900">int</span><span style="color:#999999">(</span>m <span style="color:#669900">/</span> minibatch_size<span style="color:#999999">)</span> <span style="color:#5c6370"># number of minibatches of size minibatch_size in the train set</span>
            <span style="color:#5c6370"># seed = seed + 1   #Change made by Steve Wang 博主做的改动</span>
            minibatches <span style="color:#669900">=</span> random_mini_batches<span style="color:#999999">(</span>X_train<span style="color:#999999">,</span> Y_train<span style="color:#999999">,</span> minibatch_size<span style="color:#999999">,</span> seed<span style="color:#999999">)</span>

            <span style="color:#c678dd">for</span> minibatch <span style="color:#c678dd">in</span> minibatches<span style="color:#999999">:</span>

                <span style="color:#5c6370"># Select a minibatch</span>
                <span style="color:#999999">(</span>minibatch_X<span style="color:#999999">,</span> minibatch_Y<span style="color:#999999">)</span> <span style="color:#669900">=</span> minibatch
                <span style="color:#5c6370"># IMPORTANT: The line that runs the graph on a minibatch.</span>
                <span style="color:#5c6370"># Run the session to execute the optimizer and the cost, the feedict should contain a minibatch for (X,Y).</span>
                <span style="color:#5c6370">### START CODE HERE ### (1 line)</span>
                _ <span style="color:#999999">,</span> temp_cost <span style="color:#669900">=</span> sess<span style="color:#999999">.</span>run<span style="color:#999999">(</span><span style="color:#999999">[</span>optimizer<span style="color:#999999">,</span> cost<span style="color:#999999">]</span><span style="color:#999999">,</span> feed_dict<span style="color:#669900">=</span><span style="color:#999999">{</span>X<span style="color:#999999">:</span>minibatch_X<span style="color:#999999">,</span> Y<span style="color:#999999">:</span>minibatch_Y<span style="color:#999999">}</span><span style="color:#999999">)</span>
                <span style="color:#5c6370">### END CODE HERE ###</span>
                
                minibatch_cost <span style="color:#669900">+=</span> temp_cost <span style="color:#669900">/</span> num_minibatches
                

            <span style="color:#5c6370"># Print the cost every epoch</span>
            <span style="color:#c678dd">if</span> print_cost <span style="color:#669900">==</span> <span style="color:#56b6c2">True</span> <span style="color:#669900">and</span> epoch <span style="color:#669900">%</span> <span style="color:#98c379">5</span> <span style="color:#669900">==</span> <span style="color:#98c379">0</span><span style="color:#999999">:</span>
                <span style="color:#c678dd">print</span> <span style="color:#999999">(</span><span style="color:#669900">"Cost after epoch %i: %f"</span> <span style="color:#669900">%</span> <span style="color:#999999">(</span>epoch<span style="color:#999999">,</span> minibatch_cost<span style="color:#999999">)</span><span style="color:#999999">)</span>
            <span style="color:#c678dd">if</span> print_cost <span style="color:#669900">==</span> <span style="color:#56b6c2">True</span> <span style="color:#669900">and</span> epoch <span style="color:#669900">%</span> <span style="color:#98c379">1</span> <span style="color:#669900">==</span> <span style="color:#98c379">0</span><span style="color:#999999">:</span>
                costs<span style="color:#999999">.</span>append<span style="color:#999999">(</span>minibatch_cost<span style="color:#999999">)</span>
        
        
        <span style="color:#5c6370"># plot the cost</span>
        plt<span style="color:#999999">.</span>plot<span style="color:#999999">(</span>np<span style="color:#999999">.</span>squeeze<span style="color:#999999">(</span>costs<span style="color:#999999">)</span><span style="color:#999999">)</span>
        plt<span style="color:#999999">.</span>ylabel<span style="color:#999999">(</span><span style="color:#669900">'cost'</span><span style="color:#999999">)</span>
        plt<span style="color:#999999">.</span>xlabel<span style="color:#999999">(</span><span style="color:#669900">'iterations (per tens)'</span><span style="color:#999999">)</span>
        plt<span style="color:#999999">.</span>title<span style="color:#999999">(</span><span style="color:#669900">"Learning rate ="</span> <span style="color:#669900">+</span> <span style="color:#669900">str</span><span style="color:#999999">(</span>learning_rate<span style="color:#999999">)</span><span style="color:#999999">)</span>
        plt<span style="color:#999999">.</span>show<span style="color:#999999">(</span><span style="color:#999999">)</span>

        <span style="color:#5c6370"># Calculate the correct predictions</span>
        predict_op <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>argmax<span style="color:#999999">(</span>Z3<span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span>
        correct_prediction <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>equal<span style="color:#999999">(</span>predict_op<span style="color:#999999">,</span> tf<span style="color:#999999">.</span>argmax<span style="color:#999999">(</span>Y<span style="color:#999999">,</span> <span style="color:#98c379">1</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
        
        <span style="color:#5c6370"># Calculate accuracy on the test set</span>
        accuracy <span style="color:#669900">=</span> tf<span style="color:#999999">.</span>reduce_mean<span style="color:#999999">(</span>tf<span style="color:#999999">.</span>cast<span style="color:#999999">(</span>correct_prediction<span style="color:#999999">,</span> <span style="color:#669900">"float"</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
        <span style="color:#c678dd">print</span><span style="color:#999999">(</span>accuracy<span style="color:#999999">)</span>
        train_accuracy <span style="color:#669900">=</span> accuracy<span style="color:#999999">.</span><span style="color:#669900">eval</span><span style="color:#999999">(</span><span style="color:#999999">{</span>X<span style="color:#999999">:</span> X_train<span style="color:#999999">,</span> Y<span style="color:#999999">:</span> Y_train<span style="color:#999999">}</span><span style="color:#999999">)</span>
        test_accuracy <span style="color:#669900">=</span> accuracy<span style="color:#999999">.</span><span style="color:#669900">eval</span><span style="color:#999999">(</span><span style="color:#999999">{</span>X<span style="color:#999999">:</span> X_test<span style="color:#999999">,</span> Y<span style="color:#999999">:</span> Y_test<span style="color:#999999">}</span><span style="color:#999999">)</span>
        <span style="color:#c678dd">print</span><span style="color:#999999">(</span><span style="color:#669900">"Train Accuracy:"</span><span style="color:#999999">,</span> train_accuracy<span style="color:#999999">)</span>
        <span style="color:#c678dd">print</span><span style="color:#999999">(</span><span style="color:#669900">"Test Accuracy:"</span><span style="color:#999999">,</span> test_accuracy<span style="color:#999999">)</span>
                
        <span style="color:#c678dd">return</span> train_accuracy<span style="color:#999999">,</span> test_accuracy<span style="color:#999999">,</span> parameters
</code></span>

Run the following cell to train your model for 100 epochs. Check if your cost after epoch 0 and 5 matches our output. If not, stop the cell and go back to your code!

<span style="color:#000000"><code class="language-python">_<span style="color:#999999">,</span> _<span style="color:#999999">,</span> parameters <span style="color:#669900">=</span> model<span style="color:#999999">(</span>X_train<span style="color:#999999">,</span> Y_train<span style="color:#999999">,</span> X_test<span style="color:#999999">,</span> Y_test<span style="color:#999999">)</span>
</code></span>

<span style="color:#000000"><code>Cost after epoch 0: 1.920738
Cost after epoch 5: 1.882676
Cost after epoch 10: 1.668813
Cost after epoch 15: 1.473285
Cost after epoch 20: 1.210640
Cost after epoch 25: 0.865555
Cost after epoch 30: 0.705989
Cost after epoch 35: 0.601506
Cost after epoch 40: 0.516260
Cost after epoch 45: 0.456525
Cost after epoch 50: 0.409384
Cost after epoch 55: 0.398814
Cost after epoch 60: 0.372631
Cost after epoch 65: 0.338126
Cost after epoch 70: 0.308678
Cost after epoch 75: 0.302118
Cost after epoch 80: 0.321490
Cost after epoch 85: 0.356276
Cost after epoch 90: 0.291327
Cost after epoch 95: 0.253908
</code></span>

在这里插入图片描述

<span style="color:#000000"><code>Tensor("Mean_1:0", shape=(), dtype=float32)
Train Accuracy: 0.93796295
Test Accuracy: 0.825    # dismatch the expexted output but better
</code></span>

<span style="color:#000000"><code class="language-python">fname <span style="color:#669900">=</span> <span style="color:#669900">"images/thumbs_up.jpg"</span>
image <span style="color:#669900">=</span> np<span style="color:#999999">.</span>array<span style="color:#999999">(</span>ndimage<span style="color:#999999">.</span>imread<span style="color:#999999">(</span>fname<span style="color:#999999">,</span> flatten<span style="color:#669900">=</span><span style="color:#56b6c2">False</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
my_image <span style="color:#669900">=</span> scipy<span style="color:#999999">.</span>misc<span style="color:#999999">.</span>imresize<span style="color:#999999">(</span>image<span style="color:#999999">,</span> size<span style="color:#669900">=</span><span style="color:#999999">(</span><span style="color:#98c379">64</span><span style="color:#999999">,</span><span style="color:#98c379">64</span><span style="color:#999999">)</span><span style="color:#999999">)</span>
plt<span style="color:#999999">.</span>imshow<span style="color:#999999">(</span>my_image<span style="color:#999999">)</span>
</code></span>

<span style="color:#000000"><code>C:\Users\wangh\Anaconda3\envs\tensorflow\lib\site-packages\ipykernel_launcher.py:2: DeprecationWarning: `imread` is deprecated!
`imread` is deprecated in SciPy 1.0.0.
Use ``matplotlib.pyplot.imread`` instead.
  
C:\Users\wangh\Anaconda3\envs\tensorflow\lib\site-packages\ipykernel_launcher.py:3: DeprecationWarning: `imresize` is deprecated!
`imresize` is deprecated in SciPy 1.0.0, and will be removed in 1.2.0.
Use ``skimage.transform.resize`` instead.
  This is separate from the ipykernel package so we can avoid doing imports until
</code></span>

在这里插入图片描述