Figure 2 | Scientific Reports

Figure 2

From: Affine transformations accelerate the training of physics-informed neural networks of a one-dimensional consolidation problem

Figure 2

Affine physics-informed neural network set up for one-dimensional consolidation using Biot’s theory including boundary and initial conditions with Neumann- and Dirichlet-type conditions subsumed under the boundary condition operator \({\mathscr {B}}[\cdot ]\). The losses alongside the optimization loop are depicted with red boxes. The ANN is depicted with input layer (IL), hidden layers with nonlinear activation (HL), output layer with linear activation (OL), and affine layer (AL) with the affine transformations encapsulated in orange.

Back to article page