Scribd
Upload
17 views2 pages

Assignment 5

Uploaded by

Ahmed Elkerdawy
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
17 views2 pages

Assignment 5

Uploaded by

Ahmed Elkerdawy
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Assignment 5

1. What is a Convolutional Neural Network (CNN) and what makes it suitable for image-related
tasks?

2. What are filters (kernels) in a convolutional layer, and how are they used to extract features
from images?

3. Explain the concept of local connectivity in CNNs.

4. What are the key parameters of a Convolutional Layer?

5. Why is zero-padding used in Convolutional Layers, and what impact does it have on the
spatial arrangement?

6. What is the purpose of a Pooling Layer in a ConvNet?

7. If a Pooling Layer with a filter size of 2x2 and a stride of 2 is applied to an input volume of size
28x28x64, calculate the size of the output volume.

8. Given an input volume of size 32x32x3 and a Convolutional Layer with 32 filters of size 3x3
and a stride of 1, calculate the size of the output volume.

9. For a convolutional layer with 8 filters of size 3×3 and a depth of 3, calculate the total
number of parameters, including biases.

10. For a fully-connected layer connected to an input volume of 256 × 256 × 3 how many
weights and biases are needed if there are 100 neurons in the fully-connected layer?
Challenge 1: Image Classification using Convolutional Neural Networks (ConvNet)

Objective: Build a Convolutional Neural Network (ConvNet) for image classification using only
Convolutional (Conv), Pooling (Pool), and Fully-Connected (FC) layers. The task involves
classifying images from the CIFAR-10 dataset.
Requirements:
1. Create a ConvNet architecture with a combination of Convolutional, Pooling, and Fully-
Connected layers.
2. Train the model on the CIFAR-10 training dataset.
3. Save the trained model in a .h5 file.
4. Evaluate the model on the CIFAR-10 test dataset.
Hyperparameters to Experiment With:
• Number of Convolutional Layers
• Number of Filters in Convolutional Layers
• Size of Filters in Convolutional Layers
• Pooling Type and Size
• Number of Fully-Connected Layers
• Learning Rate
• Batch Size
• Number of Epochs
Submission Instructions:
1. Submit a Notebook (.ipynb) or Python script (.py) containing the complete code.
2. Save the trained model in a file named "yourname_model.h5".
3. Report the final test accuracy achieved by your model.

You might also like