An inductive method for inexact reasoning

✍ Scribed by J. Lawry

• Publisher: Elsevier Science
• Year: 1997
• Tongue: English
• Weight: 823 KB
• Volume: 16
• Category: Article
• ISSN: 0888-613X
• DOI: 10.1016/s0888-613x(96)00135-1

✦ Synopsis

A notion of nondeterministic inference on inexact information is introduced. The model is characterized by a second order probability distribution on beliefs together with an updating procedure. Such nondeterministic inference processes naturally generate a corresponding deterministic process which essentially involves calculating the expected response of the nondeterministic process. Given that the number of propositional variables in the relevant language may vary, it is natural to consider hierarchies of inference processes, these being characterized by hierarchies of prior distributions. A number of logical principles are considered which in the presence of certain smoothness assumptions restrict the choice of hierarchies to hierarchies of symmetric Dirichlet priors where each prior is defined up to a parameter A which is constant over the whole hierarchy. A principle of maximal expected dependency between propositional variables is then introduced which restricts h to a particular value.