One hidden yet important issue for developing neural network potentials (NNPs) is the choice of training algorithm. Here we compare the performance of two popular training algorithms, the adaptive moment estimation algorithm (Adam) and the Extended Kalman Filter algorithm (EKF), using the Behler-Parrinello neural network (BPNN) and two publicly accessible datasets of liquid water [Proc. Natl. Acad. Sci. U.S.A. 2016, 113, 8368-8373 and Proc. Natl. Acad. Sci. U.S.A. 2019, 116, 1110-1115]. This is achieved by implementing EKF in TensorFlow. It is found that NNPs trained with EKF are more transferable and less sensitive to the value of the learning rate, as compared to Adam. In both cases, error metrics of the validation set do not always serve as a good indicator for the actual performance of NNPs. Instead, we show that their performance correlates well with a Fisher information based similarity measure.
翻译:开发神经网络潜力(NNPs)的一个隐藏但重要的问题是培训算法的选择。 我们在这里比较两种通用培训算法的性能,即适应性瞬间估计算法(Adam)和扩展卡尔曼过滤算法(EKFF),使用Behler-Parrinello神经网络(BPNN)和两个可公开查阅的液体水数据集[Natl. Acad. Acid. Sci. U. S. A. 2016, 113, 8368-8373和Natl. Acade. Sci. U. S. A. 2019, 116, 1110-1115]。这是通过在TensorFlow实施EKF实现的。发现,与Adam相比,接受过EKF培训的NPS对学习率价值的可转让性更低。在这两种情况下,验证组的错误度量度并非总能作为NPPs实际表现的良好指标。相反,我们表明,其性能与基于类似信息的渔业信息衡量法相干。