Hi,
On http://www.cs.ualberta.ca/~dmitri/WorldModeling
I have pointers to the comparison of DS and Bayesian approaches
as applied to the problem of range data fusion.
You might be interested in these links:
<LI> Reasoning With Uncertainty in AI (Evidence- vs
Probability- approaches)
<UL>
<LI> <A HREF="http://turing.wins.uva.nl/~fransv">
Frans Voorbraak </A>: on justification of Demspter-Shafer; DS vs
Bayesian
vs Partial Probability Theory; for robotics</LI><LI> <A
HREF="ttp://www.wins.uva.nl/research/pion/index.html">
<i>Reasoning with Uncertainty </i> project </A> </LI>
<LI>
<A
HREF="ftp://cs.nyu.edu/pub/local/hummel/papers/uncertainty.ps.Z">RobertHummel</A>,
Bayesian foundation to Demspter-Shafer method;
Relationship of DS Kalman Filtering</LI>
<LI>Dempster-Shafer Theory for Sensor Fusion in Autonomous Mobile
Robotsby
<ahref="http://www.mines.edu/fs_home/rmurphy/research/evidential.html">
Robin R. Murphy </a>: temporal fusion, sensor failure issue
<LI> Approximations for Dempster-Shafer rule:
<A HREF="http://www.dfki.de/~bauer">Mathias Bauer</a></LI>
<li> Defect in Dempster-Shafer theory (on {E,F} example: degree of
belief (which if function of accumulating weight of evidence)
doesn'tconverge to chance: Bel != Pr=t^/t):<A
HREF="http://www.cogsci.indiana.edu/farg/pwang_papers.html"> Pei Wang
</a><LI>
Thanks, Russ
Regards
Dmitry
-- Dmitry O. Gorodnichy, Ph.D. http://www.cs.ualberta.ca/~dmitri tel. (780)492-3976, (780)492-2821
> > Also, I would appreciate any ideas on how to best compare the two methods > empirically. How should I design experiments, what metrics, any other > ideas? > > Thank you in advance > Best Regards > Jim Myers, Ph.D. > ------- End of forwarded message ------- >