GMBV 2004

Guidelines for Authors

Main reasons for rejection at GMBV 2002

At the previous GMBV workshop, 7 papers were rejected for one or more of the following reasons:

5 papers were poor in content: either the work was not very innovative or the experimental results were not adequate (what is "adequate" depends on the paper: for a more theoretical paper, it could be OK to have no results on real-world images).
4 papers were not suitable for the workshop, because the work was not based on an (explicit) generative model.
2 papers were poorly written: the problem was not clearly formulated, some symbols were not defined, the terminology was confusing, etc.

Many of the accepted papers also required major revisions to the presentation.

General recommendations

A section or subsection describing the generative model(s), as distinct from the optimization algorithm(s), is recommended because it makes it easier for the reviewers to decide whether the paper is relevant to the workshop. (See also the review form.) One possible way to organize a paper is to divide it into 5 sections: Introduction, Generative Model, Methods, Results, Conclusions.

Papers accepted for the special issue will be processed differently, depending on whether they are written in LaTeX or not. Authors using LaTeX will only have to send the LaTeX files and figures to the guest editor, who will check that they compile and, when strictly necessary, make last-minute corrections with the consent of the authors.

Authors of papers not written in LaTeX will have to send the final version to the guest editor in postscript or pdf, together with all the Word/Wordperfect/whatever files: if any problems are found in the postscript/pdf, the authors will have to fix the problems and the process will be repeated to convergence. (An earlier deadline might have to be set for authors not using LaTeX.) At convergence, the guest editor will send the Word/Wordperfect/whatever files to the publishers without checking them: the publisher might find more problems later. Papers of large size (several Mbytes) require a large amount of storage and time to download, uncompress, and print: please consider using jpeg-compressed images and jpeg2ps to wrap jpeg images into eps format. Of course, plots should not be jpeg-compressed.

Important details

All symbols used in the paper should be defined before they are first used: write

"Define the energy E, the mass m, and the speed of light c. Then we have: E = mc²"

rather than:

"We have: E = mc², where E is the energy, m is the mass, and c is the speed of light."

(The latter is still better than no definition at all, of course.)

All plots must have both axes clearly labelled with the variables being plotted and the units of measurement. The axis labels should be large enough to be readable, i.e. no smaller than the font size of the figure legend; the same goes for the numbers on the axis tics. To save time for the reviewers, papers will be checked for unlabelled plots before the papers are sent out for review.

A list of all model/method parameters and their values in the numerical experiments is strongly encouraged.

Nitpicking

Please refer to equations and figures in the style "Eq. (1), Fig. 1", for consistency across papers.

When writing fractions in the text, please write x/y rather than \frac{x}{y}.

No iterative subscripting or superscripting please: do not write x_{y_t}, write instead x_y(t) or x(y_t) or x_y,t or whatever is suitable. Also, please write exp(argument) rather than e^argument unless the argument is very simple.

Review form

Look here if you want to know how your paper will be assessed.