论文标题
旋转模棱两可的卷积的相关性用于预测分子特性
Relevance of Rotationally Equivariant Convolutions for Predicting Molecular Properties
论文作者
论文摘要
模棱两可的神经网络（ENNS）是嵌入$ \ mathbb {r}^3 $中的图形神经网络，非常适合预测分子特性。 ENN库E3NN具有可自定义的卷积，可以设计仅取决于点之间或角特征之间的距离，从而使它们分别旋转不变或均衡。本文研究了通过\ texttt {e3nn}和QM9数据集的消融研究直接研究分子属性预测的角度依赖性的实际值。我们发现，对于固定的网络深度和参数计数，添加角特征将测试误差降低23％。同时，增加网络深度的平均值仅减少了4％的测试误差，这意味着旋转等效的层相对有效。我们介绍了偶极时刻准确性提高的解释，该目标从引入角特征中受益最大。
Equivariant neural networks (ENNs) are graph neural networks embedded in $\mathbb{R}^3$ and are well suited for predicting molecular properties. The ENN library e3nn has customizable convolutions, which can be designed to depend only on distances between points, or also on angular features, making them rotationally invariant, or equivariant, respectively. This paper studies the practical value of including angular dependencies for molecular property prediction directly via an ablation study with \texttt{e3nn} and the QM9 data set. We find that, for fixed network depth and parameter count, adding angular features decreased test error by an average of 23%. Meanwhile, increasing network depth decreased test error by only 4% on average, implying that rotationally equivariant layers are comparatively parameter efficient. We present an explanation of the accuracy improvement on the dipole moment, the target which benefited most from the introduction of angular features.
