Automatic feature engineering for catalyst design using small data without prior knowledge of target catalysis

TAKAHASHI Keisuke

Department of Chemistry

Jan. 16, 2024

図 1 開発技術を用いた触媒性能の予測。（上部）開発技術が異なる触媒反応に対して高 精度な回帰を与えること、（下部）従来技術（元素組成のみ, 元素組成+特徴量選択, 特徴 量付与のみ）と比較して極めて高い精度を与えることが示されています。1～3 次は合成 特徴量の次数を指し、次数が増加するほど、より複雑な特徴量をより大量に生成します。

Joint press release (in Japanese)

Abstract
The empirical aspect of descriptor design in catalyst informatics, particularly when confronted with limited data, necessitates adequate prior knowledge for delving into unknown territories, thus presenting a logical contradiction. This study introduces a technique for automatic feature engineering (AFE) that works on small catalyst datasets, without reliance on specific assumptions or pre-existing knowledge about the target catalysis when designing descriptors and building machine-learning models. This technique generates….

Read the full article on Communications Chemistry

Article Information:
Taniike, T., Fujiwara, A., Nakanowatari, S., García-Escobar, F ., Takahashi, K
 Automatic feature engineering for catalyst design using small data without prior knowledge of target catalysis. Commun Chem7, 11 (2024).
DOI: 10.1038/s42004-023-01086-y