Microcomputers in Civil Engineering - Journal of Computer-Aided Civil and Infrastructure Engineering: Special Issue on Machine Learning, vol. 12, no. 4, 1997.

Edited by Yoram Reich

Content [In alphabetical order of authors]:

• Biedermann, J. (1997) Representing Design Knowledge with Neural Networks.
  Abstract: Neural networks have been used in a number of civil engineering applications because of their ability to implicitly learn an input-output relationships. Typically the applications involve deriving an input-output relationships for problems that may be too complex to mathematically model, computationally, expensive or difficult to solve using the traditional procedural computing approach. Heuristic design knowledge used by structural engineers when performing structural design often falls in the latter category of being difficult to represent procedurally. Neural networks have been investigated for the representation of heuristic design knowledge and the results of this investigation and the lessons learned regarding neural network training is presented.
  
  
• Manevitz, L., Yousef, M. and Givoli, D. (1997) Finite Element Mesh Generation Using Self-Organized Neural Networks.
  Abstract: Neural networks are applied to the problem of mesh placement for the finite element method. When the finite element method is used to numerically solve a partial differential equation with boundary conditions over a domain; the domain must be divided into "elements". The precise placement of the nodes of the elements has a major affect on the accuracy of the numerical method. In this paper the self-organizing algorithm of Kohonen is adapted to solve the problem of automatically assigning (in a near-optimal way) coordinates from a two dimensional domain to a given topological grid (or mesh) of nodes in order to apply the finite element method effectively when solving a partial differential equation with boundary conditions over that domain.
  One novelty of the method is the interweaving of versions of the Kohonen algorithm in different dimensions simultaneously in order to handle the boundary of the domain properly.
  Our method allows for the use of arbitrary types of two dimensional elements (in particular, quadrilaterals or mixed as opposed to just triangular) and for varying desired densities over the domain. (Thus more elements can be automatically placed near "areas of interest".)
  The methods and experiments developed here are for two-dimensional domains but seem naturally extendible to higher dimensional problems. The method uses a mixture of both one and two-dimensional versions of the Kohonen algorithm, with an improvement suggested by Tabakman and Exman, and further adapted to the particular problem here. Experimental results comparing this algorithm with a well known two-dimensional grid generating system (PLTMG) are presented.
  
  
• Pompe, P. P. M. and Feelders (1997) Using Machine Learning and Statistics to Predict Corporate Bankruptcy of Construction Companies: A Comparative Study.
  Abstract: Recent literature strongly suggests that machine learning approaches to classification outperform "classical" statistical methods. We make comparison between the performance of linear discriminant analysis, classification trees and neural networks in predicting corporate bankruptcy. Linear discriminant analysis represents the "classical" statistical approach to classification, whereas classification trees and neural networks represent artificial intelligence approaches. A proper statistical design is used to be able to test whether observed differences in predictive performance are statistically significant. The dataset consists of a collection of 576 annual reports from Belgian construction companies. We used stratified 10-fold cross-validation on the training set to choose "good" parameter values for the different methods. The test set is used to obtain an unbiased estimate of the true prediction error. Using rigorous statistical testing we can not conclude that one method clearly outperforms the other methods.
  
  
• Reich, Y. (1997), Machine Learning Techniques for Civil Engineering Problems.
  Abstract: The growing volume of information databases presents opportunities for advanced data analysis techniques from machine learning (ML) research. Practical applications of ML are very different from theoretical or empirical studies, involving organizational and human aspects, and various other constraints. Despite the importance of applied ML, little has been discussed in the general ML literature on this topic. In order to remedy this situation, we studied practical applications of ML and developed a proposal for a seven-steps process that can guide practical applications of ML in engineering. The process is illustrated by relevant applications of ML in civil engineering. This illustration shows that the potential of ML has only begun to be explored, but also cautions that in order to be successful, the application process must carefully address the issues related to the seven-step process.
  
  
• Sanchez, M., Cortes, U., R.-Roda, I., Poch, M. and Lafuente, J. (1997) Learning and Adaptation in Wastewater Treatment Plants through Case-Based Reasoning.
  Abstract: This paper discusses a case-based reasoning approach to model the specific or experiential knowledge directly coming from wastewater treatment plants (WWTP) actual operation, within the overall supervision task of a plant. A concrete implementation is detailed: case structure, case library organization, retrieving algorithm, matching function and learning techniques. Starting from some initial cases (learning-by-observation), the system evolves adapting (learning-by-own-experience) its experiential knowledge from the actual operation of the concrete WWTP under control.The result is a more accurate supervisory system. Recording previous experiences -cases- occurred in the system helps to solve new similar or related situations in a plant with less effort than other methods that start from the scratch to build-up new solutions. Moreover, the continuous execution of the system enhances its adaptation to new situations that could appear in the WWTP.
  
  
• Zhang, J. and Yang, J. (1997), An Application of Instance-Based Learning to Highway Accident Frequency Prediction.
  Abstract: Accurate predictions of highway accident frequency may help traffic engineers design and test solutions for the improvements of highway safety. However, accident frequency prediction is by no means an easy task due to the large number of factors for accident occurrence and complicated interactions among them. Many studies have been conducted to uncover the relationship between roadway environment and corresponding accident frequencies. These studies used statistical approaches such as linear regression. The actual relationship has not been approximated with acceptable certainty because it usually coincided with the mathematical models assumed by the researchers. This paper describes an application of Instance-Based Learning (IBL) to highway accident frequency predictions. This application involves predictions of numeric values rather than classifications. We designed and implemented an IBL system, called IBPA, for numeric value predictions and applied this system to highway accident frequency predictions. The dataset used contains accident data from the main Utah highways for a five year period (1988-1992). It includes 1077 instances. Experimental results shows that IBL methods are applicable to highway accident predictions and compared favorably to linear regression.
  
  

Last modified by Yoram Reich on February 10, 1996