Guidelines for paper critiques

Critique Guidelines

The main goal of this class is to understand basic techniques that can be used for modeling complex shapes and their interactions in time. Some of the papers were selected because they give interesting partial solutions, while others were chosen because they suggest interesting approaches or mathematical techniques.

[The text below is adapted from a page posted by Carlo Tomasi for his Computer Vision Tracking course offered in the spring quarter '98-'99.]

Critiques must be in PDF format. The URL of your critique should be e-mailed to the instructor one week before your presentation, so they can be handed out in class a week ahead of the discussion. For instance, the critique for the February 25 presentation should be e-mailed by February 18.

In the following, I list some of the items that ought to appear in your paper critiques. However, it is up to you to add to these elements anything you deem useful, interesting, or fun.

	Put the complete and correct citation for the paper on the critique.
	Summarize the paper in a few sentences at the beginning of your critique. This should be your own summary.
	Elaborate on your synopsis with a brief summary of the paper, highlighting what is new, what is old, and what is important. Sometimes definitions or brief explanations of difficul t or technical aspects of the paper are appropriate.
	State in what way this paper contributes to our understanding of tracking. For instance, does the paper describe a mathematical technique from a different field? If so, how can we use it for tracking, and what modifications are we likely to have to make? Are these modifications trivial, or are major breakthroughs needed? If the paper describes a vision tracking system, what is useful about it? How can we improve on it?
	What assumptions need to be made for the techniques described in the paper to work? List both the assumptions explicitly made by the authors and any others you may think of yourse lf. Give specific examples in which the techniques would and would not work.
	Is the content of the paper good? This question is delicate, and has several facets. When reading a technical paper, we are sometimes dissatisfied with the quality or generality o f the results, the level of mathematical treatment, or the style of the presentation. However, critical reading does not mean finding fault with the paper. On the contrary, it means finding the gold nugget in it, no matter how deeply it may be buried. A brilliant idea poorly presented is still a brilliant idea. If an algorithm requires many assumptions, it may just mean that the author was more careful or more honest than others in making those assumptions explicit. A paper that overlaps 90 percent with something else may still contain 10 percent new material. Please be positive in your comments, but do point out flaws.
	Do the experiments in the paper, if any, back up the claims? Are the results interpreted correctly? Are the experiments representative of the situations for which the proposed me thod is likely to be used? Are the experiments convincing?
	Does the paper integrate knowledge from other fields? What background did the authors come from, what are the weaknesses of the method, what are its strengths, both for its inten ded goals and for tracking in computer vision?
	Is the paper well written? Is its structure satisfactory? Is it clear? Is proper reference made to other papers in the field? Is the mathematics useful, or does its complexity hi de lack of ideas or get in the way of comprehension? These questions are secondary given the purposes of our class, but they will help you understand how to write better papers yourself.
	How does the paper compare or contrast with other papers that the class has read or is reading for that week? Can concepts read somewhere else integrate or strengthen this paper? Are there contradictions or incompatibilities with other papers we have read? Were papers or books mentioned in the bibliography that we should be aware of for our tracking work?
	In your critique, provide a set of issues or questions to lead off a discussion. This can be done by asking a series of questions about the paper, or by advocating strong opinions for or against a given method. This item is important for our class discussions. Writing your main points in your critique is a way for both you and all of us to think about the topic ahead of time.
	Finally, here is a link to example critiques written by former students in Carlo's class.