Fwd: Re: [UAI] Definition of Bayesian network

From: Peter McBurney (P.J.McBurney@csc.liv.ac.uk)
Date: Mon Aug 06 2001 - 21:14:18 PDT

  • Next message: Goebel, Kai F (CRD): "[UAI] CFP: SPIE Fusion Conference Abstracts Due 8/20/01"

    In view of Kathy's comment:

    "However, it's worth noting that Phil Dawid, Vladimir Vovk, and Glenn
    Shafer have been working on an alternate formulation of probability
    ("prequential games"), based on agents engaged in sequential games.
    (Shafer and Vovk just published a book which they announced on this
    list.)"

    in response to Kevin's earlier comment:

    "The product and sum rules of probability theory give us the proof theory of
    our logic. Set-theoretic probability theory gives us the model theory for our
    logic."

    it is worth noting that prequential games may be viewed as a game-theoretic
    semantics (model theory), in the sense of logician Jaako Hintikka (although his
    ideas may be traced to the work of C. S. Peirce a century ago). In this
    approach, a statement in some logical language is assigned the value "true" if
    a particular player in an associated game has a winning strategy, and "false"
    otherwise. Game-semantics have recently found application in theoretical
    computer science, as a semantics for computer programming languages. (Work by
    e.g. Abramsky, McCusker, Hyland, Schalk.) The approach has also recently been
    applied (unwittingly) to model and predict financial markets; see the web-pages
    of David Lamper at Oxford University:

             http://www.maths.ox.ac.uk/~lamper/index2.html

    Regarding alternative semantics for probability, UAI list members may
    be interested in a nice recent book by philosopher Donald Gillies
    (although it does not discuss prequential games semantics):

    D. Gillies (2000): "Philosophical Theories of Probability". London, UK:
    Routledge.

    This compares and contrasts logical, subjectivist, frequentist and propensity
    theories of probability in an accessible and historically-minded fashion.
    After comparing these, Gillies argues for an eclectic approach, in particular
    for a subjectivist philosophy of probability in the social sciences and
    a non-subjectivist philosophy in the natural sciences.

     

    - -- Peter

    ****************************************************************
                                                              
      Peter McBurney
      Agent Applications, Research and Technology (Agent ART) Group
      Department of Computer Science
      University of Liverpool
      Liverpool L69 7ZF
      U.K.
                                                              
      Tel: + 44 151 794 6768
      Email: P.J.McBurney@csc.liv.ac.uk
      Web page: www.csc.liv.ac.uk/~peter/
                                                                                                                         
    ****************************************************************



    This archive was generated by hypermail 2b29 : Mon Aug 06 2001 - 21:21:24 PDT