BriefGPT.xyz
Oct, 2019
应用于分子性质预测的图神经网络上的多任务学习
Multitask Learning On Graph Neural Networks Applied To Molecular Property Predictions
HTML
PDF
Fabio Capela, Vincent Nouchi, Ruud Van Deursen, Igor V. Tetko, Guillaume Godin
TL;DR
提出了一种基于图神经网络模型的最新、高效的多任务预测方法，结果表明，多任务学习可以提高模型性能，特别是数据点较少的数据集可以不需要数据增强，便能获得较好的效果，并且能显著减小模型的方差。
Abstract
Prediction of
molecular properties
, including
physico-chemical properties
, is a challenging task in chemistry. Herein we present a new state-of-the-art multitask prediction method based on existing graph neural n
→