MML mixture models of heterogeneous poisson processes with uniform outliers for bridge deterioration

Authors:
T. Maheswaran;J. G. Sanjayan;David L. Dowe;Peter J. Tan
Affiliations:
Previously Department of Civil Engineering, Monash University when the research presented in this paper was carried out, Vic;Department of Civil Engineering, Monash University, Clayton, Vic, Australia;School of Computer Science and Software Engineering, Monash University, Clayton, Vic, Australia;School of Computer Science and Software Engineering, Monash University, Clayton, Vic, Australia
Venue:
AI'06 Proceedings of the 19th Australian joint conference on Artificial Intelligence: advances in Artificial Intelligence
Year:
2006

Citing 5
Cited 0

MML clustering of multi-state, Poisson, vonMises circular and Gaussian distributions

Statistics and Computing
Finding Cutpoints in Noisy Binary Sequences - A Revised Empirical Evaluation

AI '99 Proceedings of the 12th Australian Joint Conference on Artificial Intelligence: Advanced Topics in Artificial Intelligence
Minimum Message Length Grouping of Ordered Data

ALT '00 Proceedings of the 11th International Conference on Algorithmic Learning Theory
Statistical and Inductive Inference by Minimum Message Length (Information Science and Statistics)

Statistical and Inductive Inference by Minimum Message Length (Information Science and Statistics)
MML inference of oblique decision trees

AI'04 Proceedings of the 17th Australian joint conference on Advances in Artificial Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

Effectiveness of maintenance programs of existing concrete bridges is highly dependent on the accuracy of the deterioration parameters utilised in the asset management models of the bridge assets. In this paper, bridge deterioration is modelled using non-homogenous Poisson processes, since deterioration of reinforced concrete bridges involves multiple processes. Minimum Message Length (MML) is used to infer the parameters for the model. MML is a statistically invariant Bayesian point estimation technique that is statistically consistent and efficient. In this paper, a method is demonstrated estimate the decay-rates in non-homogeneous Poisson processes using MML inference. The application of methodology is illustrated using bridge inspection data from road authorities. Bridge inspection data are well known for their high level of scatter. An effective and rational MML-based methodology to weed out the outliers is presented as part of the inference.