This project compares two convolutional neural network architectures on the Fashion MNIST dataset:
- SimpleCNN: A basic CNN with 2 convolutional layers.
- DeeperCNN: A deeper CNN with 4 convolutional layers.
The goal is to evaluate and visualize the marginal performance differences between these models using accuracy, loss, ROC curves, and probability-based visualizations.
How to Run
- Create New AIS Project
- Install dependencies:
pip install -r requirements.txt3. Run the notebook: Open
fashion_mnist_comparison.ipynb
in Jupyter Lab and execute all cells sequentially.
Resources:
https://github.com/HPInc/aistudio-samples/tree/main/hackathon-sample-projects/hackathon-retail-MNIST
Walkthrough video: Hackathon-FashionMNIST-retail.mp4