C1_W3_Lab_1_improving_accuracy_using_convolutions

Course Q&A TensorFlow Developer Professional Introduction to TF for Artificial Intelligence ...
1 
import matplotlib.pyplot as plt
f, axarr = plt.subplots(3,7)
FIRST_IMAGE=0
SECOND_IMAGE=4
THIRD_IMAGE=26
CONVOLUTION_NUMBER = 1
from tensorflow.keras import models
layer_outputs = [layer.output for layer in model.layers]
activation_model = tf.keras.models.Model(inputs = model.input, outputs = layer_outputs)
for x in range(0,4):
  f1 = activation_model.predict(test_images[FIRST_IMAGE].reshape(1, 28, 28, 1))[x]
  axarr[0,x].imshow(f1[0, : , :, CONVOLUTION_NUMBER], cmap='inferno')
  axarr[0,x].grid(False)
  f2 = activation_model.predict(test_images[SECOND_IMAGE].reshape(1, 28, 28, 1))[x]
  axarr[1,x].imshow(f2[0, : , :, CONVOLUTION_NUMBER], cmap='inferno')
  axarr[1,x].grid(False)
  f3 = activation_model.predict(test_images[THIRD_IMAGE].reshape(1, 28, 28, 1))[x]
  axarr[2,x].imshow(f3[0, : , :, CONVOLUTION_NUMBER], cmap='inferno')
  axarr[2,x].grid(False)

In the above code, f1 contains the result of each output layer right? Am I understanding it correct?

2

Hi Varunm,
yes.

You initialized it from
“layer_outputs = [layer.output for layer in model.layers]”
by using List comprehension.

The difference between F1,F2,F3 are the images that been used.
For Example:
“test_images[FIRST_IMAGE]” → it mean you refer to index [FIRST_IMAGE=0] from from test_image list

3

Hey ryokusnadi, appreciate if you can help me clarify these questions below. Thanks!

what does it mean by changing the CONVOLUTION_NUMBER, is there an upper bound for this parameter, what the range we are allowed to pick up for this typical case?

The code is no longer working even I changed the CONVOLUTION_NUMBER back to 1 after I experimented it with a big number, i.e. 20. Why does this happen?

Screen Shot 2021-09-12 at 9.34.57 PM
Screen Shot 2021-09-12 at 9.34.57 PM1682×1208 359 KB

4

Hi CCCC,
“is there an upper bound for this parameter, what the range we are allowed to pick up for this typical case?”
→ I Don’t think so

The code is no longer working even I changed the CONVOLUTION_NUMBER back to 1 after I experimented it with a big number, i.e. 20. Why does this happen?
→ I think you are using wrong subplots, try change it into : f, axarr = plt.subplots(3,4)

5

hi ryokusnadi

i have a question for this part.

We have 4 picture in every row (corresponding every f),

so the first picture is output of the first convolution layer (feature map 1), the second picture is output of the second convolution layer (feature map 1), the third picture is output of the first dense anh the final picture is output of the remaining dense, right?

Can you explain for me?

6

Hi,
We have 4 picture in every row (corresponding to every f)
→ Do you mean “(1, 28, 28, 1)” for 4 picture?
you can think (1, 28, 28, 1) as 1 image 28 by 28 pixels in greyscale(last index of 1)

We only have 3 images for this case, and it was used for indexing.
FIRST_IMAGE=0
SECOND_IMAGE=4
THIRD_IMAGE=26

7

im sorry, i mean, with the first image with index = 0, transformed throught 3 layer conv2d and 1 layer dense. with every layer conv2d, will mention feature map 1.
example: first colunm in the below picture.
image