CNN architecture for image classification in TensorFlow - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to add a convolutional layer with 32 filters and a 3x3 kernel.

TensorFlow

model.add(tf.keras.layers.Conv2D([1], (3, 3), activation='relu', input_shape=(28, 28, 1)))

Drag options to blanks, or click blank then click option'

A32

B128

C16

D64

Attempts:

3 left

💡 Hint

Common Mistakes

Using too few filters like 16 may limit feature extraction.

Using too many filters like 128 can increase computation unnecessarily.

The first convolutional layer typically uses 32 filters to extract features from the input images.

2fill in blank

medium

Complete the code to add a max pooling layer with pool size 2x2.

TensorFlow

model.add(tf.keras.layers.MaxPooling2D(pool_size=[1]))

Drag options to blanks, or click blank then click option'

A(3, 3)

B(1, 1)

C(2, 2)

D(4, 4)

Attempts:

3 left

💡 Hint

Common Mistakes

Using (1, 1) does not reduce the size.

Using (3, 3) or larger may reduce size too much.

MaxPooling2D with pool size (2, 2) reduces the spatial dimensions by half, helping to reduce computation and control overfitting.

3fill in blank

hard

Fix the error in the code to flatten the output before the dense layer.

TensorFlow

model.add(tf.keras.layers.[1]())

Drag options to blanks, or click blank then click option'

ADense

BDropout

CConv2D

DFlatten

Attempts:

3 left

💡 Hint

Common Mistakes

Using Dense instead of Flatten causes shape errors.

Using Conv2D or Dropout here is incorrect.

Flatten layer converts the 2D feature maps into a 1D vector so it can be fed into dense layers.

4fill in blank

hard

Fill both blanks to add a dense layer with 128 units and ReLU activation.

TensorFlow

model.add(tf.keras.layers.Dense([1], activation=[2]))

Drag options to blanks, or click blank then click option'

A128

B'sigmoid'

C'relu'

D64

Attempts:

3 left

💡 Hint

Common Mistakes

Using sigmoid activation in hidden layers can slow learning.

Choosing too few units like 64 may limit model capacity.

A dense layer with 128 units and ReLU activation helps the model learn complex patterns.

5fill in blank

hard

Fill all three blanks to compile the model with Adam optimizer, sparse categorical crossentropy loss, and accuracy metric.

TensorFlow

model.compile(optimizer=[1], loss=[2], metrics=[[3]])

Drag options to blanks, or click blank then click option'

A'sgd'

B'sparse_categorical_crossentropy'

C'accuracy'

D'adam'

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'sgd' optimizer may require tuning learning rate.

Using 'categorical_crossentropy' instead of sparse version causes errors if labels are integers.

Adam optimizer is popular for training CNNs. Sparse categorical crossentropy is used for multi-class classification with integer labels. Accuracy measures how often predictions are correct.