inr_classification

INR classification

The following commands assume that you are executing them from the current directory experiments/inr_classification. If you are in the root of the repository, please navigate to the experiments/inr_classification directory:

cd experiments/inr_classification

Activate the conda environment:

conda activate neural-graphs

Setup

Download the data

For INR classification, we use MNIST and Fashion MNIST. The datasets are available here.

• MNIST INRs
• Fashion MNIST INRs

Please download the data and place it in dataset/mnist-inrs and dataset/fmnist_inrs, respectively. If you want to use a different path, please change the following commands accordingly, or symlink your dataset path to the default ones.

MNIST

wget "https://www.dropbox.com/sh/56pakaxe58z29mq/AABrctdu2U65jGYr2WQRzmMna/mnist-inrs.zip?dl=0" -O mnist-inrs.zip &&
  mkdir -p dataset/mnist-inrs &&
  unzip -q mnist-inrs.zip -d dataset &&
  rm mnist-inrs.zip

Fashion MNIST

wget "https://www.dropbox.com/sh/56pakaxe58z29mq/AAAssoHq719OmSHSKKTiKKHGa/fmnist_inrs.zip?dl=0" -O fmnist_inrs.zip &&
  mkdir -p dataset/fmnist_inrs &&
  unzip -q fmnist_inrs.zip -d dataset &&
  rm fmnist_inrs.zip

Data preprocessing

We have already performed the data preprocessing required for MNIST and Fashion MNIST and provide the files within the repository. The preprocessing generates the data splits and the dataset statistics. These correspond to the files dataset/mnist_splits.json and dataset/mnist_statistics.pth for MNIST, and dataset/fmnist_splits.json and dataset/fmnist_statistics.pth for Fashion MNIST.

However, if you want to use different directories for your experiments, you have to run the scripts that follow, or simply symlink your paths to the default ones.

MNIST

First, create the data split using:

python dataset/generate_mnist_data_splits.py \
  --data-path mnist-inrs --save-path . --name mnist_splits.json

This will create a json file dataset/mnist_splits.json. Note that the --data-path and --save-path arguments should be set relatively to the dataset directory.

Next, compute the dataset (INRs) statistics:

python dataset/compute_mnist_statistics.py \
  --data-path . --save-path . \
  --splits-path mnist_splits.json --statistics-path mnist_statistics.pth

This will create dataset/mnist_statistics.pth object. Again, --data-path and --save-path should be set relatively to the dataset directory.

Fashion MNIST

Fashion MNIST requires a slightly different preprocessing. First, prepare the data splits using:

python dataset/preprocess_fmnist.py \
  --data-path fmnist_inrs/splits.json --save-path . --name fmnist_splits.json

Next, compute the dataset statistics:

python dataset/compute_fmnist_statistics.py \
  --data-path . --save-path . \
  --splits-path fmnist_splits.json --statistics-path fmnist_statistics.pth

This will create dataset/fmnist_statistics.pth object.

Run the experiment

Now for the fun part! 🚀 To train and evaluate a Neural Graph Transformer (NG-T) model on the MNIST dataset, run the following command:

python main.py model=rtransformer data=mnist

Make sure to check the model configuration in configs/model/rtransformer.yaml and the data configuration in configs/data/mnist.yaml. If you used different paths for the data, you can either overwrite the default paths in configs/data/mnist.yaml or pass the paths as arguments to the command:

python main.py model=rtransformer data=mnist \
  data.dataset_dir=<your-dataset-dir> data.splits_path=<your-splits-path> \
  data.statistics_path=<your-statistics-path>

Training a different model or using a different dataset is as simple as changing the model and data arguments! For example, you can train and evaluate a Neural Graph Graph Neural Network (NG-GNN) on Fashion MNIST using the following command:

python main.py model=pna data=fmnist

Run experiments with scripts

You can also run the experiments using the scripts provided in the scripts directory. For example, to train and evaluate a Neural Graph Transformer (NG-T) model on the MNIST dataset, run the following command:

./scripts/mnist_cls_rt.sh

This script will run the experiment for 3 different seeds.