China 
Africa 
Explore chapters and articles related to this topic
Deep Learning
Published in Peter Wlodarczak, Machine Learning and its Applications, 2019
A ReLU is a ramp function and graphically it looks as shown in Figure 8.1. Despite its simplicity, it can handle non-linear and interaction effects surprisingly well. Non-linear effects simply means that the relationship between the independent variables and the predictions is not linear, in other words, the slope of the graph is not constant. Interaction effects means that a variable a affects a prediction differently depending on variable b. For instance, if we want to predict the risk of cardiovascular problems in patients based on body weight, we also need to know the height of the patients.
Non-linear analysis of vertically loaded pile groups
Published in W.F. Van Impe, W. Haegeman, Deep Foundations on Bored and Auger Piles, 2020
The linear analysis described above can be extended to include non-linear effects by generalising an idea first proposed by Poulos & Davis (1968). In the present analysis, the boundary element equations used for the linear response are solved incrementally while enforcing the conditions of yield, equilibrium and compatibility at the pile-soil interface.
Value of information-based risk and fatigue management for offshore structures
Published in Journal of Structural Integrity and Maintenance, 2020
A. Agusta, B.J. Leira, S. Thöns
For this example, the system failure probability is approximated by considering only overload and single fatigue failure sequence (see Equation (4)). Furthermore, it is assumed that uncertainties of the strength and geometry parameters of members are negligible which makes a deterministic variable. is calculated by removing one component at a time and determining the ultimate base shear by performing pushover analysis with USFOS. Non-linear effects such as non-linear material behavior, large deformations and imperfections are explicitly modeled. In this study, 200 pushover analyses are performed to calculate for every possible realizations of .