Showing 1–2 of 2 results for author: Stillman, J

Uncertainty and Incompleteness

Authors: Piero P. Bonissone, David A. Cyrluk, James W. Goodwin, Jonathan Stillman

Abstract: Two major difficulties in using default logics are their intractability and the problem of selecting among multiple extensions. We propose an approach to these problems based on integrating nommonotonic reasoning with plausible reasoning based on triangular norms. A previously proposed system for reasoning with uncertainty (RUM) performs uncertain monotonic inferences on an acyclic graph. We ha… ▽ More Two major difficulties in using default logics are their intractability and the problem of selecting among multiple extensions. We propose an approach to these problems based on integrating nommonotonic reasoning with plausible reasoning based on triangular norms. A previously proposed system for reasoning with uncertainty (RUM) performs uncertain monotonic inferences on an acyclic graph. We have extended RUM to allow nommonotonic inferences and cycles within nonmonotonic rules. By restricting the size and complexity of the nommonotonic cycles we can still perform efficient inferences. Uncertainty measures provide a basis for deciding among multiple defaults. Different algorithms and heuristics for finding the optimal defaults are discussed. △ Less

Submitted 27 March, 2013; originally announced April 2013.

Comments: Appears in Proceedings of the Fifth Conference on Uncertainty in Artificial Intelligence (UAI1989)

Report number: UAI-P-1989-PG-34-45

On Heuristics for Finding Loop Cutsets in Multiply-Connected Belief Networks

Authors: Jonathan Stillman

Abstract: We introduce a new heuristic algorithm for the problem of finding minimum size loop cutsets in multiply connected belief networks. We compare this algorithm to that proposed in [Suemmondt and Cooper, 1988]. We provide lower bounds on the performance of these algorithms with respect to one another and with respect to optimal. We demonstrate that no heuristic algorithm for this problem cam be guar… ▽ More We introduce a new heuristic algorithm for the problem of finding minimum size loop cutsets in multiply connected belief networks. We compare this algorithm to that proposed in [Suemmondt and Cooper, 1988]. We provide lower bounds on the performance of these algorithms with respect to one another and with respect to optimal. We demonstrate that no heuristic algorithm for this problem cam be guaranteed to produce loop cutsets within a constant difference from optimal. We discuss experimental results based on randomly generated networks, and discuss future work and open questions. △ Less

Submitted 27 March, 2013; originally announced April 2013.

Comments: Appears in Proceedings of the Sixth Conference on Uncertainty in Artificial Intelligence (UAI1990)

Report number: UAI-P-1990-PG-265-272