Bayesian neural networks and Gaussian processes in identification of concrete properties
Journal Title: Computer Assisted Methods in Engineering and Science - Year 2011, Vol 18, Issue 4
Abstract
This paper gives a concise overview of concrete properties prediction using advanced nonlinear regression approach and Bayesian inference. Feed-forward layered neural network (FLNN) with Markov chain Monte Carlo stochastic sampling and Gaussian process (GP) with maximum likelihood hyperparameters estimation are introduced and compared. An empirical assessment of these two models using two benchmark problems are presented. Results on these benchmark datasets show that Bayesian neural networks and Gaussian processes have rather similar prediction accuracy and are superior in comparison to linear regression model.
Authors and Affiliations
Marek Słoński
Keywords
Related Articles
- EP ID EP74037
- DOI -
- Views 166
- Downloads 0
The analysis of long-term borehole heat exchanger system exploitation
The key issue in designing borehole heat exchangers (BHE) is the long-term performance of the ground source heat pump (GSHP) systems. The performance directly reflects economic profitability and depends on a large number...
Problems of the equilibrium of a rigid body and mechanical systems. (Received in the final form August 12, 2009)
In this article one of the greatest generalized methods for establishing the equilibrium equations of a rigid body and the set of rigid bodies is proposed. It is related to six equations of moments of force about six the...
Optimization of functionally gradient materials in valve design under cyclic thermal and mechanical loading
The robust and simple optimization method of functionally graded material (FGM) for combined cyclic thermal and mechanical loading with application to valve design is proposed. The optimization procedure starts from the...
Architecture of iterative solvers for hp-adaptive finite element codes
We present a layered architecture for iterative solvers of linear equations, designed to allow for easy integration with existing hp-adaptive FEM codes. We discuss interfaces between a solver and an external FEM code and...
Solving thermoelasticity problems by means of Trefftz functions. (Received in the final form January 15, 2010)
The paper presents a new method of approximate solving of the two- and three-dimensional thermoelasticity problems in a finite body. The method presented here can be used for solving direct and inverse problems as well....