Fashion MNIST Classification

This repository contains code for training a neural network to classify images from the Fashion MNIST dataset.

Requirements

all the needed requirements can be downloaded using this line:

pip install -r requirements.txt

Files

train.py: Main training script
fashion_mnist_classification_nn_pytorch.py: Contains the FashionMNISTDataModule, FashionMNISTClassifier, and FashionMNISTClassifier2 classes
config.yaml: Configuration file

Models

There are two different models available:

Weak Model (FashionMNISTClassifier): A simple feedforward neural network with two hidden layers (128 and 64 units) and dropout.
Powerful Model (FashionMNISTClassifier2): A deeper feedforward neural network with four hidden layers (392, 196, 98, and 49 units) and dropout.

Usage

To train the neural network, run the following command:\

python3 train.py optimizer.name=<optimizer_name> optimizer.lr=<learning_rate> batch_size=<batch_size> epochs=<epochs> classifier=<model_type>, example=<showing_example_or_not>

optimizer.name: Can be "adam" or "sgd".
optimizer.lr: Learning rate for the optimizer, can be any positive float
batch_size: Batch size for training and validation, can be any positive integer
epochs: Number of epochs to train for, can be any positive integer
classifier: Model type to use for training, can be "weak" or "powerful"
example : the trained model will predict a single random image from the dataset.

Example

To train the model using the Adam optimizer with a learning rate of 0.001, batch size of 64, for 20 epochs and the first classifier type, run the following command:

python3 train.py optimizer.name=adam optimizer.lr=0.001 batch_size=64 epochs=20 classifier=powerful example=True

Results

The training and validation losses and accuracies will be logged after each epoch. After training, the model will be saved in the trained_models directory with the current timestamp and configuration details.

For any issues or questions, feel free to create an issue on the repository.