Causal Inductive Reasoning: A New Paradigm for
Data-Driven Qualitative Simulation
of Continuous-Time Dynamical Systems
Keywords
- Inductive Reasoning
- Qualitative Modeling
- Fuzzy Systems
- Prediction
- Redundant Reasoning
Abstract
Inductive reasoning attempts to induce future behavioral patterns of a system
or time series from observation of the behavioral patterns of their past. No
assumption is made about the underlying model structure. Inductive reasoning
represents a paradigm of pure behavioral modeling and simulation of systems.
Like all other qualitative approaches to reasoning, inductive reasoning fights
a constant battle of generality versus specificity, a battle against ambiguity
stemming from uncertainty. Causal inductive reasoning tries to win this battle
by introducing redundancy into the reasoning process. Causal inductive
reasoning can be compared to the president of a company who works with multiple
independently operating marketing analysts. All of his or her analysts have to
make predictions under uncertainty. If more than one of them comes to the same
conclusion, the president will be more likely to heed their advice.
Interested in reading the
full paper?
(17 pages, 139,177 bytes, pdf)
Homepage
Last modified: November 29, 2006 -- © François Cellier