Scribd
Upload
11 views2 pages

Fashion MNIST Image Classification Assignment

This assignment involves developing a neural network model for classifying images from the Fashion MNIST dataset, which contains 70,000 grayscale images of fashion products across 10 categories. Students will preprocess the data, design and train a convolutional neural network, and evaluate its performance, culminating in a report detailing their methodology and insights. The submission is due on March 13, 2024, and must include code and a PDF report hosted on GitHub.

Uploaded by

jeronojeddy
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
11 views2 pages

Fashion MNIST Image Classification Assignment

This assignment involves developing a neural network model for classifying images from the Fashion MNIST dataset, which contains 70,000 grayscale images of fashion products across 10 categories. Students will preprocess the data, design and train a convolutional neural network, and evaluate its performance, culminating in a report detailing their methodology and insights. The submission is due on March 13, 2024, and must include code and a PDF report hosted on GitHub.

Uploaded by

jeronojeddy
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

Deep Learning Assignment (In groups of three): Image Classification with

Fashion MNIST Dataset

Submission Date: 13th March 2024

Overview

In this assignment, you will work with the Fashion MNIST dataset, a modern replacement for the
classic MNIST dataset. It includes 28x28 grayscale images of 70,000 fashion products
categorized into 10 types. The dataset is split into 60,000 training images and 10,000 test images.
Your task is to develop a neural network model that can accurately classify these images into
their respective categories.
Objectives
• To understand and apply the concepts of neural networks and deep learning.
• To gain experience with preprocessing image data.
• To experiment with neural network architectures for image classification.
• To evaluate and analyze the performance of your model.
Dataset Categories
The Fashion MNIST dataset includes the following categories:
0. T-shirt/top
1. Trouser
2. Pullover
3. Dress
4. Coat
5. Sandal
6. Shirt
7. Sneaker
8. Bag
9. Ankle boot
Tasks
1. Data Preprocessing
• Load the Fashion MNIST dataset.
• Normalize the image data to range between 0 and 1.
• Reshape the data to fit the model input requirements.
• Convert the labels into one-hot vectors.
2. Model Development
• Design a convolutional neural network (CNN) suitable for classifying the images.
• Compile your model with an appropriate optimizer, loss function, and metrics.
3. Model Training
• Train your model on the training data.
• Validate your model using the test data.
4. Evaluation and Analysis
• Evaluate the model's performance on the test dataset.
• Plot the training and validation accuracy and loss over epochs.
• Display sample images with their predicted and actual labels.
5. Report
• Write a report summarizing your methodology, model architecture, training
process, and results.
• Discuss any challenges you faced and how you overcame them.
• Provide insights on how the model's performance could be improved.
Submission Requirements
• Via GitHub. Use Google Colab notebook or Python script containing all the code for data
preprocessing, model building, training, and evaluation.
• A PDF report covers all the points mentioned in the "Report" section.
Evaluation Criteria
• Correctness and completeness of the data preprocessing steps.
• Suitability and creativity in the neural network model design.
• Accuracy of the model on the test data.
• Depth of analysis in the report and the quality of insights provided.
Deadline
Submit your completed assignment (code and report) by [insert deadline here].
Resources
• TensorFlow and Keras documentation for reference and guidance.
• Matplotlib for data visualization.
Notes
• Make sure to comment on your code for clarity.
• Experiment with different model architectures and hyperparameters to achieve the best
performance.
• Ensure your report is concise, well-organized, and reflects a thorough understanding of
the task.

You might also like