MoonNote

Comparison of Atom Representations in Graph Neural Networks for Molecular Property Prediction 본문

Study/Paper Review

Comparison of Atom Representations in Graph Neural Networks for Molecular Property Prediction

Kisung Moon 2021. 12. 14. 16:35

3줄 요약

- 분자의 물성을 예측하기 위해 다양한 feature들이 사용되고 있음

- 12가지 조합의 atom representation과 선행 연구에서 사용한 representation을 사용하여 성능을 비교함

- 표현이 많아질수록 성능이 좋아지는 경향이 있지만 의외로 aromaticity, ring, formal charge 정보는 제거하는 것이 좋았고 heavy neighbours 및 hydrogen 을 추가하면 좋은 성능이 나타나는 경향이 있었다.

 

Comment

GCN으로만 실험을 진행하였기 때문에 분자에 다양한 GNNs(GAT, GIN, SchNet, SGCN 등) 들로 실험을 하여 결과를 살펴볼 필요가 있다.

Comments