Machine Learning - CNN, DT, NN implementation

A Machine Learning university project, where we create our own implementations of a neural network and a decision tree. We also used PyTorch to create a CNN. Our task was to classify different types of clothing based on an image of the item.

Important Details

• group work: we worked in pairs
• the submission was code + report, which then was discussed individually in the exam
• dataset: https://www.kaggle.com/datasets/zalando-research/fashionmnist

What I would do different?

• I would add one more layer to the nn
• I would use ReLU as a main activation function, but softmax as the last activation function
• I would use MiniBatch GD in a different way