Abstract:
Given the description of a physical system in one of several forms (a set of constraints, Bayesian network etc.) and a set of observations made, the task of model-based diagnosis is to find a suitable assignment to the modes of behavior of individual components (this notion can also be extended to handle transitions and dynamic systems [Kurien and Nayak, 2000]. Many formalisms have been proposed in the past to characterize diagnoses and systems. These include consistency-based diagnosis, fault models, abduction, combinatorial optimization, Bayesian model selection etc. Different approaches are apparently well suited for different applications and representational forms in which the system description is available. In this paper, we provide a unifying theme behind all these approaches based on the notion of model counting. By doing this, we are able to provide a universal characterization of diagnoses that is independent of the representational form of the system description. We also show how the shortcomings of previous approaches (mostly associated with their inability to reason about different elements of knowledge like probabilities and constraints) are removed in our framework. Finally, we report on the computational tractability of diagnosis-algorithms based on model counting. 1
Citations
|
660
|
A theory of diagnosis from first principles
– Reiter
- 1987
|
|
519
|
Diagnosing multiple faults
– Kleer, Williams
- 1987
|
|
205
|
Probabilistic horn abduction and bayesian networks
– Poole
- 1993
|
|
157
|
A Model-based Approach to Reactive Self-Configuring Systems
– Williams, Nayak
- 1996
|
|
153
|
Diagnosing with behavioral modes
– Kleer, Williams
- 1989
|
|
143
|
Characterizing diagnoses and systems
– Kleer, Mackworth, et al.
- 1987
|
|
116
|
A spectrum of logical definitions of model-based diagnosis
– Console, Torasso
- 1991
|
|
75
|
Physical Negation - Integrating Fault Models into the General Diagnostic Engine
– Struss, Dressler
- 1989
|
|
73
|
Monte-Carlo approximation algorithms for enumeration problems
– Karp, Luby, et al.
- 1989
|
|
73
|
Explanation in the situation calculus
– Shanahan
- 1993
|
|
48
|
A methodology for using a default and abductive reasoning system
– Poole
- 1990
|
|
46
|
and Johan de Kleer; Building Problem Solvers
– Forbus
- 1993
|
|
42
|
to the Future for Consistency-based Trajectory Tracking
– Kurien, Nayak, et al.
- 2000
|
|
42
|
Diagnosis of Continuous Valued Systems in Transient Operating Regions
– Mosterman, Biswas
- 1999
|
|
41
|
A theory of diagnosis for incomplete causal models
– Console, Dup´re, et al.
- 1989
|
|
35
|
Explanatory diagnosis: conjecturing actions to explain observations
– McIlraith
- 1998
|
|
31
|
Fast context switching in real-time propositional reasoning
– Nayak, Williams
- 1998
|
|
29
|
Constraint networks. Encyclopedia of Artificial Intelligence
– Dechter
- 1992
|
|
28
|
An Assumption Based TMS
– deKleer
- 1986
|
|
25
|
Counting models using connected components
– Pehoushek
|
|
22
|
On the tractability of counting theory models and its application to belief revision and truth maintenance
– Darwiche
- 2001
|
|
20
|
F.: Analysis of notions of diagnosis
– Lucas
- 1998
|
|
15
|
Representing diagnosis knowledge
– Poole
- 1994
|
|
9
|
Diagnosis as a Trial: The Alibi Principle
– Raiman
- 1989
|
|
5
|
Bayesian Model-Based Diagnosis
– Lucas
- 2001
|
|
4
|
Model-Based Diagnosis and Probabilistic Assumption-Based Reasoning
– Kohlas, Anrig, et al.
- 1998
|
|
4
|
Extensions to ATMS-based Diagnosis
– Struss
- 1988
|
|
2
|
QCBFS: Leveraging Qualitative Knowledge in Simulation-Based Diagnosis
– Kumar
- 1985
|
|
1
|
An Information-Theoretic Characterization of Abstraction in Diagnosis and Hypothesis Selection
– Kumar
- 2002
|
