Some Thesis Comments
 
 I think you are quite near a very good thesis
 Here are some comments -- not in order
 1) Airline problems are solved by LP (Linear Programming)
    Can you use LP in your problem
    Why use your methods not LP in airline?
 2) Can Graph Coloring be used as a preprocessor?
 3) Discuss importance of robust solutions i.e. still works after 5% of
    students change
 4) Mention concept of physical optimization and the papers of Rose
    (to find these, grep http://www.new-npac.org/users/fox/homepage/gcfstuff/cv.html
    I just updated some papers with him)
 5) Why is Energy (SA) different from Energy (MFA)> I can see that MFA is more
    sensitive than SA, but I would have guessed that what is best for MFA is
    also best for SA
 6) I suggest some sort of hierarchial diagram categorizing differen methods
    you use e.g. set of questions
    is there an (artificial) energy function
    are monte carlo or deterministic methods used etc.
 7) You sometimes talk a lot about something before you define it.
    e.g. MFA and Hopfield properties are discussed before they are
    described
    This is confusing except to expert reader
    You mention "mapping" a lot but don't define
 8) You should discuss more clearly why a neural net can be used in optimization