2023年度 人工知能学会全国大会（第37回）
2023年6月6日〜6月9日熊本城ホール(熊本県熊本市) ＋ オンライン
2023年度 人工知能学会全国大会（第37回）
2023年6月6日〜6月9日熊本城ホール(熊本県熊本市) ＋ オンライン
[1U3-IS-2a-02]Improving the efficiency of training physics-informed neural networks using active learning
〇Yuri Aikawa1, Naonori Ueda2, Toshiyuki Tanaka1(1. Univ. of Kyoto, 2. NTT Communication Science Laboratories)
[[Online, Regular]]
PINN is a PDE solver realized as a neural network by incorporating the PDEs to be satisfied into the network as physical constraints. In this study, focusing on how to select the collocation points, we propose an active learning method to improve the efficiency of PINN learning. The proposed method
uses variational inference based on dropout learning to evaluate the uncertainty of the solution estimate by PINN and defines an acquisition function for active learning based on the uncertainty. Then, by probabilistically sampling collocation points using the acquisition function, a reasonable solution can be obtained faster than random sampling. We demonstrate the effectiveness of the method using Burgers’ equation and the convection equation. We also show experimentally that the choice of the collocation points can affect the loss function, the fitting of initial and boundary conditions, and the sensible balance of PDE constraints.
uses variational inference based on dropout learning to evaluate the uncertainty of the solution estimate by PINN and defines an acquisition function for active learning based on the uncertainty. Then, by probabilistically sampling collocation points using the acquisition function, a reasonable solution can be obtained faster than random sampling. We demonstrate the effectiveness of the method using Burgers’ equation and the convection equation. We also show experimentally that the choice of the collocation points can affect the loss function, the fitting of initial and boundary conditions, and the sensible balance of PDE constraints.