No abstract

Bayesian network (BN) models can be used to predict the level of fault injection in the different phases of the software development process. This paper describes the techniques used by Motorola to develop BN fault prediction models using product, usage and process information. It also includes a procedure created to validate and calibrate the BN models. Statistical techniques are applied that provide a measurement of the quality of the predictions made by the models, and directions on how the models could be improved. The validation method described in this paper considers two alternative ways to produce an improved network. The first is based mainly on modifying the ranges of the existing nodes, adding interdependencies between them, and varying the weight values associated with each of the nodes that are used as inputs to the intermediate nodes of the BN model. The second possibility uses linear regression and principal component analysis to build the intermediate and output nodes of the network. The paper closes with some encouraging results, and outlines a number of unresolved questions. Copyright © 2006 John Wiley & Sons, Ltd.

employing the Multilayer perceptron (MLP) as knowledge representation form is very dependent on the degree of complexity that the structure of the application 's domain has, Different mathematical and/or statistical techniques have been developed to subtract the maximum amount of information of this type from an available sample of the operating space associated to the task of interest. In the context of MLP it has been used to decide on the form the diferent intervening parameters of the network and/or related learning algorithm (LA) should have. This paper provides an overview of the processes that have been defined to generate network applications using the MLP model, giving particular attention to those based on the dynamic creation of a network's architecture through the application of different techniques for subtracting information about the operating domain in which the training set is subsumed.

I Document-based management of requirements can be effective f o r single site, narrowly scoped projects. However, we firmly believe that it is essential to move to a database-centered way of managing requirements i f platform-based product development and enterprise wide requirements management is to be fully supported. This paper introduces an approach to the specification of system families that is database centered. It outlines the experience of producing a database of requirements for the TV product. The schema was built using RTM, a requirements management tool supplied by Integrated Chipware. It constitutes a practical example of the advantages that can be gained by changing from the conventional document based requirements management process. It also shows the difficulties that need to be dealt with when mapping from a document-based to a database centered product specification. Events

No abstract