UofT-DSI | deep_learning - Assignment 1

[ruokun-ai]

What changes are you trying to make? (e.g. Adding or removing code, refactoring existing code, adding reports)

I added code to build and evaluate image classification models using the Fashion MNIST dataset. The changes include loading and inspecting the dataset, creating a baseline linear model, building a CNN model, running controlled experiments with different filter sizes and dropout regularization, and training a final model based on the best experiment results.

What did you learn from the changes you have made?

I learned how to prepare image data for Keras models, including normalization, reshaping, and one-hot encoding. I also learned how CNNs improve image classification performance by learning spatial features from images, compared with a simple baseline model that only uses flattened pixel values.

Was there another approach you were thinking about making? If so, what approach(es) were you thinking of?

Another approach would be to use a deeper CNN with more convolutional layers or to test additional hyperparameters such as kernel size, learning rate, batch size, and dropout rate. I kept the model simple to match the assignment requirements and make the experiments easier to compare.

Were there any challenges? If so, what issue(s) did you face? How did you overcome it?

One challenge was making sure the label format matched the loss function. For one-hot encoded labels, I used categorical_crossentropy. I also needed to create a new model instance for each experiment so that weights from previous models were not reused. I resolved these issues by checking the model setup and keeping the experiment code consistent.

How were these changes tested?

The changes were tested by running the notebook from top to bottom. I evaluated the baseline model, the simple CNN model, the filter-size experiments, the dropout experiment, and the final model using test loss and test accuracy. I also checked the outputs and tables to compare model performance.

A reference to a related issue in your repository (if applicable)

NA

Checklist

• I can confirm that my changes are working as intended

[github-actions]

Hello, thank you for your contribution. If you are a participant, please close this pull request and open it in your own forked repository instead of here. Please read the instructions on your onboarding Assignment Submission Guide more carefully. If you are not a participant, please give us up to 72 hours to review your PR. Alternatively, you can reach out to us directly to expedite the review process.