Author name cluster

Stefan

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

2 papers

1 author row

IJCAI Conference 2005 Conference Paper

On Solution Correspondences in Answer-Set Programming

Thomas Eiter
Hans Tompits
Stefan

We introduce a general framework for specifying program correspondence under the answer-set semantics. The framework allows to define different kinds of equivalence notions, including previously defined notions like strong and uniform equivalence, in which programs are extended with rules from a given context, and correspondence is determined by means of a binary relation. In particular, refined equivalence notions based on projected answer sets can be defined within this framework, where not all parts of an answer set are of relevance. We study general characterizations of inclusion and equivalence problems, introducing novel semantical structures. Furthermore, we deal with the issue of determining counterexamples for a given correspondence problem, and we analyze the computational complexity of correspondence checking.

PDF

IJCAI Conference 2005 Conference Paper

Training without data: Knowledge Insertion into RBF Neural Networks

Ken McGarry
Stefan

A major problem when developing neural networks for machine diagnostics situations is that no data or very little data is available for training on fault conditions. However, the domain expert often has a good idea of what to expect in terms of input and output parameter values. If the expert can express these relationships in the form of rules, this would provide a resource too valuable to ignore. Fuzzy logic is used to handle the imprecision and vagueness of natural language and provides this additional advantage to a system. This paper investigates the development of a novel knowledge insertion algorithm that explores the benefits of prestructuring RBF neural networks by using prior fuzzy domain knowledge and previous training experiences. Pre-structuring is accomplished by using fuzzy rules gained from a domain expert and using them to modify existing Radial Basis Function (RBF) networks. The benefits and novel achievements of this work enable RBF neural networks to be trained without actual data but to rely on input to output mappings defined through expert knowledge.

PDF