/physml

Utilize machine learning and deep learning in atomization energy prediction

Primary LanguageJupyter Notebook

Atomization Energies Prediction with QM7 Dataset

Suggestions are always welcome!

Introduction

This repository contains the recipe for predicting atomization energies of small molecules in QM7 dataset, using various machine learning algorithms and deep learning models, which are MLPs and GNNs.

Details about those algorithms and models can be found in the report.

Prequisites

To run the code, you first need to setup the repo by running at the root folder:

pip install -e .

Usage

Details about how to train and evaluate the models can be found in 2 folders:

  • ml: for machine learning algorithms.
  • nn: for deep learning models.

Results

Please check the report for more details.

References

  1. Learning Invariant Representations of Molecules for Atomization Energy Prediction
  2. Neural Message Passing for Quantum Chemistry
  3. Assessment and Validation of Machine Learning Methods for Predicting Molecular Atomization Energies
  4. Gated Graph Sequence Neural Networks