Re: Direction for a Ph.D research

> I am starting a Ph.D. about  Building Intellegent Expert system for
> diagnosis of medical images. I know there are two schools for doing
> that: the numeric school and the symbolic school.

You have to ask yourself whether you will be a) putting human
knowledge into the system (symbolic approaches are more appropriate),
or b) acquiring knowledge inductively from symptoms (symbolic
approaches are better) or from images (numerical approaches are less
constrained). Whatever you do, you better make sure that you properly
deal with uncertainty: you need to deal with statistics, either
Fisherian or Bayesian. Once you deal with uncertainty, there is no
major distinction between symbolic and numeric approaches. Bayesian
approaches have a neat way of including human knowledge into the model
in the form of priors.

There are many journals and conferences. Ask your thesis advisor to
give you directions and tell you which conferences he attends and
which journals he reads. You should start your research in the field
of your advisor, otherwise you may get lost.

Aleks

[ comp.ai is moderated.  To submit, just post and be patient, or if ]
[ that fails mail your article to <[EMAIL PROTECTED]>, and ]
[ ask your news administrator to fix the problems with your system. ]