Skip to content

Latest commit

 

History

History
89 lines (67 loc) · 4.84 KB

README.md

File metadata and controls

89 lines (67 loc) · 4.84 KB
Raw

Graph Sum-Product Networks

This repository contains the code for the following paper: Papež M, Rektoris M, Šmídl V, Pevný T. GraphSPNs: Sum-Product Networks Benefit From Canonical Orderings. In The 7th Workshop on Tractable Probabilistic Modeling 2024.

An illustration of the key principles behind various GraphSPNs. (a) The exact permutation invariance first computes $p^{\text{spn}}_{m,n}$ for all $n!$ permutations and then averages the results. However, this approach is computationally infeasible even for very small graphs. Therefore, we investigate the following different principles to ensure the permutation invariance. (b) The $k$-ary permutation invariance approximates the exact invariance (a) by computing $p^{\text{spn}}_{k,n}$ for all ways to choose $k$-node sub-graphs from the $n$-node original graph, $G$, without repetition and with order, and then averaging the $M={n!}/{(n-k)!}\leq n!$ results. (c) The random sampling approximates the exact invariance (a) by computing the average only for $N\lt n!$ permutations of $G$. (d) The sorting approach is also exact, but it simplifies the target data distribution by first imposing the same canonical ordering of $G$ and then computing $p^{\text{spn}}_{m,n}$.

@inproceedings{papez2024graphspns,
 title={Graph{SPN}s: Sum-Product Networks Benefit From Canonical Orderings},
 author={Milan Pape\v{z} and Martin Rektoris and V\'{a}clav \v{S}m\'{i}dl and Tom\'{a}\v{s} Pevn\'{y}},
 booktitle={The 7th Workshop on Tractable Probabilistic Modeling},
 year={2024},
 url={https://openreview.net/forum?id=NKdl7FHlFK}
}

1. Install

Clone this repository.

git clone https://github.com/mlnpapez/GraphSPN GraphSPN

Go to the GraphSPN directory.

cd GraphSPN

Set up the environment.

conda create --name graphspn python==3.8.5

source activate graphspn

pip install torch==2.1.2 --index-url https://download.pytorch.org/whl/cu121
pip install rdkit==2022.9.5
pip install tqdm==4.65.0
pip install pandas==2.0.3
pip install pylatex==1.4.2

The implementation relies on the Einsum networks. Run mkdir einsum and fill einsum with the content of https://github.com/cambridge-mlg/EinsumNetworks/tree/master/src/EinsumNetwork.

2. Preprocess

The following command will download and preprocess two versions of the QM9 dataset. qm9_sort.pt contains molecules with the canonical ordering of the atoms. qm9_perm.pt contains molecules with a random ordering of the atoms.

python -m utils.datasets

3. Train

config/qm9/ contains JSON files with the GraphSPN variants' hyper-parameters. Change the hyper-parameters based on your preferences and then run the following command.

python -m train

It will train all the GraphSPN variants (or only the selected ones if you change the list of names in train.py).

The resulting models will be stored in results/training/model_checkpoint/, and the corresponding illustrations of unconditional molecule generation, along with the metrics assessing the performance of the models, will be stored in results/training/model_evaluation/.

Unconditional samples of molecular graphs from the sort variant of GraphSPNs (graphspn_zero_sort).

4. Gridsearch

gridsearch_hyperpars.py contains hyper-parameter grids to find a suitable architecture for the GraphSPN variants. Change the hyper-parameter grids based on your preferences, and then run the following command.

nohup python -m gridsearch > gridsearch.log &

This command will run the script in the background, submitting jobs to your SLURM cluster. The resulting models, metrics, and output logs will be stored in results/gridsearch/model_checkpoint/, results/gridsearch/model_evaluation/, and results/gridsearch/model_outputlogs/, respectively.

To reproduce the results in the paper (Table 1), keep the current settings in gridsearch_hyperpars.py. Then, after completing all the SLURM jobs, run the following command.

python -m gridsearch_evaluate

It will produce Table 1 from the paper (both in the .pdf and .tex formats).

4. Conditional Molecule Generation

Run the following command to generate new molecules conditionally on a known molecule.

python -m conditional_sampling

To impose a known structure of the generated molecules, change patt_smls in conditional_sampling.py. Similarly, to select a model from which to generate the samples, change model_path.

Conditional samples of molecular graphs from the sort variant of GraphSPNs (graphspn_zero_sort). The known part of a molecule is highlighted in blue.