Paper outline:

1. Abstract (1 or at most 2 paragraphs): What is it?

   Abstracts are meant to interest the reader and draw him/her into reading the paper.

   • A summary of the paper.

   • Designed for someone in the field. That is, they (the readers) know about a bit about AI/neuralnets/DT/GAs

   • Never use abbreviations in the abstract

   • Contents of abstract:

     1. What is the Problem

     2. What is your method of attack

     3. Are the Results promising?

2. Introduction (a few paragraphs): What is it? Someone is interested in your work because they read the abstract. Now, provide :

   1. Motivation

      • What is the problem (expanded from abstract) and why is it interesting (application areas)

      • Have other people found it interesting?

      • Why use <your-method> on this problem?

   2. Background

      • Who has done work in this area before? Do a web search. For each relevant paper summarize the work, how it is related to yours, and how it is different. Sometimes the major similarities and differences can be summarized at the end of this part.

      • Cite the paper properly. You need to look at the citations in the example paper for this. Quinlan must be cited for decision trees, Holland and Goldberg for Genetic Algorithms, and Rumelhart and Mclelland for NNs.

   3. Short description of Methodology:

      Your audience is some one in CS who may not know much about your technique or your application area.

      • a paragraph on Technique

      • a paragraph or two on application area is appropriate

   4. Summarize your results. Don't say how you got these results, just what they were. Notice that you will be repeating your results several times in a paper. This is normal, repetition drives (others') learning.

   5. Outline or structure of the rest of the paper.

   The idea is to get the reader to ask how did you get the results you got? To find out they better read the rest of the paper, or, if they are more familiar with the topic, to go to the appropriate section.

3. Methodology: For someone who may not know much about DTs/NNs/GAs or problem area. Flesh out the material in the introduction. Methodology may require more than one section, usually for representation or encoding.

   1. What is a DT/GA/NN? The algorithm, the theory.

   2. What is your problem and how do you use a GA/DT/NN on your particular problem?

   3. How did you process the training data? How did you change your the canonical DT/NN/GA to work on this problem and why?
   4. Representation, operators, heuristics

4. Results and Analysis (may be two sections)

   • What did you do?

   • Why?

   • and was the outcome what you expected?

   • Why or why not.

   This section usually has a lot of figures and tables. For algorithms that are non-deterministic (GAs/NNs) you must provide statistically significant results. Simply running a GA/NN once on a problem and reporting results is usually useless. However, having consistently (across many random seeds) good/bad performance is a good indication of result quality. Note, high quality does not equate to high performance.

   • The format is usually: For each Idea I had

     • The neat idea i had of applying DT/GAs/NNs to my problem led me to do expect X.

     • When I tried it, here's what happened and why

5. Conclusions and Future Work:

   Summarize the results (again) and the CONSEQUENCES of your results. Consequences means (among other things) impact on your field.

   • What questions does your work raise?

   • Your suggestions for how to answer those questions OR Conjectures (informed guess as to what the answer is) and some intuition to support your conjectures.

   • Most people in your field will read the abstract, results, and this section.

   • Be very careful about what you say. If you say something that does not feel right or is completely wrong you will completely destroy the validity of your entire work.

   • At the same time, if you don't make the right conclusions and raise the right questions, you will once again destroy the validity of the entire body of work you did.

   • this is one of the IMPORTANT sections, don't think that it will be easy. Spend time on it.

6. Bibliography/References: Any style is ok but it must be complete