5月学内共同研究促進トーク

5月の学内共同研究促進トークは、計算システムズ生物学研究室の小野直亮先生にご講演いただきました。

タイトルと概要は以下になります。

====================

小野直亮 准教授 (計算システムズ生物学研究室)

TITLE:  Embedding Natural Organic Compounds into Latent Space by Autoencoder Using Molecular Hypergraph Grammar

Abstract:

We propose a model for embedding natural organic compounds into a latent space using a molecular hypergraph grammar-based autoencoder.
The goal is to capture the inherent structural features and chemical properties of organic compounds in a compact and meaningful representation. The encoder maps the input compounds, represented as hypergraphs, into a lower-dimensional latent space.
To evaluate the effectiveness of our approach, we construct a classifier of the organic compounds into manually annotated hierarchical categories such as Terpenoids-Triterpenoids-Limonoids, etc.
This embedding enables various downstream applications, such as compound clustering, similarity searching, and property prediction.