The First instructional Conference on Machine Learning

Reviews

Reviewers should complete the following form for each paper.

Rate the paper on each of the attributes below (1=worst, 5=best). Justify your choice with one or more sentences. Be kind. Negative reviews are much easier to accept if they sound constructive and enthusiastic instead of nasty, dismissive, or condescending. Find something good to say, especially if you need to say some strong negative things as well.

1 2 3 4 5: Clarity of Writing

It's very important that readers not be distracted by sloppy writing or confusing English. Documents should be spell checked before being submitted. (1=needs substantial rewriting, 3=a few mistakes, 5=very well edited.)  
 
 
 
 
 
 
 
 
 

1 2 3 4 5: Completeness

In research, it is very rare for an idea to work the first time. It usually takes multiple experiments to really understand what's going on. Often, papers are written that are mostly an idea with little experimental work to flesh it out. (1=very preliminary, 3=just enough work to support the main idea, 5=very complete story.)  
 
 
 
 
 
 
 
 
 

1 2 3 4 5: Originality

There's always something new in a paper, but it's usually pretty minor. A conference paper should contain an original idea or a new way of looking at an existing idea. (1=old news, 3=a new idea, 5=a revolutionary idea.)  
 
 
 
 
 
 
 
 
 

1 2 3 4 5: Experimental Work

All of the papers in iCML are experimental and there are levels of quality of experimental work. Does the experiment answer a specific question? Is the measured quantity interesting? Are confidence intervals given? Are there multiple experiments to help prevent misinterpretation of results? (1=sloppy, 3=solid, 5=flawless.)  
 
 
 
 
 
 
 
 
 

1 2 3 4 5: Overall Recommendation

Ultimately, your job as a reviewer is to provide a recommendation to the senior program committee (SPC) as to whether or not the paper should be accepted for presentation in the conference. If all of the reviewers give the same recommendation, the SPC will probably go along with it. If there is disagreement, the SPC member might ask you to clarify your comments to try to come up with well reasoned decision. (1=reject, 2=lean to reject, 3=could go either way, 4=lean to accept, 5=accept.)

Reasons to accept:  
 
 
 
 
Reasons to reject:  
 
 
 
 

Yes No: Nominate for best paper award?

Usually there is a committee who decides on a best paper award, but the committee wants to know which papers to look at most closely.