[UAI] Robert Example

From: zadeh@cs.berkeley.edu
Date: Tue Mar 27 2001 - 22:00:20 PST

  • Next message: Kevin Korb: "[UAI] Research Fellow in Computational Data Analysis and Modelling (2 positions)"

    The Robert Example (corrected)

          The Robert example is an instance of a class of uncertainty-centered
    problems which lie beyond the reach of classical probability theory and
    Bayesian techniques. However, the questions posed in the Robert example
    fall within the reach of a perception-based generalization of probability
    theory. The Robert example relates to everyday commonsense reasoning. It is
    intended to draw attention to a major shortcoming of classical probability
    theory and decision analysis the inability to operate on perception-based
    information.

          More specifically, assume that I am considering calling Robert
    at home at 6:30 pm. The question is: Should I call him
    person-to-person or station-to-station, given that the calls cost a
    and b, respectively.

          The decision-relevant information is the probability, P, that
    Robert is home at 6:30 pm. Thus, the question is: What is P?

          There are three versions of the information which I have -- in
    order of increasing difficulty of analysis.

    Version l. My perception is that Robert usually returns from work at about
    6:00 pm.

    Version 2. My perceptions are:
    (a) Robert usually leaves his office at about 5:30 pm; and
    (b) it usually takes him about 30 minutes to get home.

    Version 3. My perceptions are:
    (a) Robert usually leaves his office at about 5:30 pm; and
    (b) the travel time depends on the time when he leaves office.

    More specifically:

    when Robert leaves at about 5:l5 pm, the travel time is usually about
    25 minutes.

    when Robert leaves at about 5:30 pm, the travel time is usually about
    30 minutes.

    when Robert leaves at about 5:45 pm, the travel time is usually about
    35 minutes.

    An additional question: What is the earliest time at which the
    probability that Robert is home is high?

    It is understood that in the Robert example, terms such as high,
    usually, about 6:00 pm, etc., do not have a crisp meaning.

    =====
    Lotfi A. Zadeh
    Professor in the Graduate School, Computer Science Division, Department of
    EECS,
    University of California, Berkeley, CA 94720-1776.
    Director, Berkeley Initiative in Soft Computing (BISC)
    Tel (office): (510) 642-4959, Fax (office): (510) 642-1712
    Tel (home): (510) 526-2569, Fax (home): (510) 526-2433, (510) 526-5181
    zadeh@cs.berkeley.edu,
    http://www.cs.berkeley.edu/People/Faculty/Homepages/zadeh.html



    This archive was generated by hypermail 2b29 : Tue Mar 27 2001 - 22:06:59 PST