CAP 6635: Artificial Intelligence
Project Announcement
2022 Spring
Term project proposal guideline [Final term project report due by May 6. Firm deadline]
The goal of the term project is for students to practice Bayesian network and Naïve bayes classification
for text classification. The final outcomes of the project need to be turned in as a formal technical report.
Students can choose to use a term project report to substitute the final exam, or they can choose to
take the final exam and submit a simplified term project report (Please refer to technical report
instructions for details).
Term Project and Final Exam
All students must participate the term project and submit the term project report. Follow plot below to
determine which routine (term project, term project reports, final exam) you will choose.
Do you want to use term project to
substitute final exam?
Can form a team with 3 members maximum
Must follow “Research Report Instruction”
No Final Exam
Report Due: May 6
Can only work solo.
Follow “Course Report Instruction”
Final Exam on May 4
Report Due: May 6
Term Project
All students
YES NO
Students have options to (1) use term project report to substituent final exam, or (2) turn in a short
version of the term project report and participate final exam:
Students using term project to substitute final exam: Students can choose to substitute the final exam
using term project report. If this is the case, the term project report will contribute 10+15=25 points to
your final grade. The “Research Report” needs to have 4,000 words minimum including motivation and
description of the research problems, technical solutions, validations, conclusion, and reference etc.
(The “Research Report Instruction” posted in the Canvas includes all details). I will assist each student
(or each team) to polish the report and try to find a suitable venue to publish the report, if possible.
Students participating in final exam: For students who still want to take the final exam, the term project
eport will contribute 10 points to your final grade. Your report shall have 1,000 words minimum,
including statement of the research problem, designs, and validations. (The “Course Report Instruction”
posted in the Canvas includes all details). The “standardized term project” is posted as an option for
students participating in final exam. But students are welcome to propose their own term project topic,
even if you do NOT want to use template project to substitute the final exam.
Students choosing to participate final exam CANNOT team up to work on the term project
Naïve Bayes for Text Classification and Transfer Learning
The goal of the standardized project for students to exercise on designing machine learning methods for
text classification.
Due to limited time, students are required to follow the naïve bayes for text classification to complete
the term project. Please refer to [NB for Text Classification and transfer learning [Notebook, html]]
notebook for code example on how to design naïve bayes classification for text classification, and
transfer learning. To complete term assignment, your project must achieve following functionality.
1. Must use email.zip dataset (which include normal and spam emails) as training set, and validate
the model (Naïve Bayes based models or additional model) performance on a training vs. test
splitting set.
2. Must evaluate how stop words impact on the classification results
3. Must evaluate how does an NB model trained from email dataset performance on another SMS
dataset (The dataset has two files: sms.csv include another short message dataset with 5572
short messages, and labels.csv include label information of each message: normal vs. spam)
4. Must have a discussion on how to improve the performance of a model trained from email.zip
on SMS dataset (a typical transfer learning scenario).
For students who wish to use term project to submit the final exam, the report must meet following
criteria
https:
canvas.fau.edu/courses/115657/modules/items/3250127
• You must have a design and implementation of your method in order to outperform a simple NB
model trained from email.zip to classify messages in SMS dataset
• You can only use at most 5% of messages in SMS dataset for training (the rest 95% of samples
must be excluded from the training phase) and the remaining messages in SMS dataset being
used as test set.
• You can use all messages in email.zip for training.
• You can use additional dictionary or stop words provided by a third party. For example, short
messages have short a
eviations, you can use those a
eviation in your model.
• You must use 10 time repetitions or 10 fold-cross validation to validate and report your results.
CAP 6635: Artificial Intelligence (2022 Spring)
Research Report Instruction
(For Students Substituting Final Exam) (10 points (term) + 15 Points (final))
Due date [May 6 2022, Firm]
This instruction only applies to students who intend to use a longer version of the term
project report (i.e., a research report) to substitute the final exam. If you DO wish to
participate in the final exam, please follow the “Course Report Instruction (For students
participating in final exam)” in the Canvas.
The grading of the term project report is based on the following criteria.
1. Overall [3 pts]: You should organize your report in IEEE format, with 4,000 words minimum. Please
note that table/figure do not count towards the word limitation. You can use IEEE word or Latex
temperate from the following URL
a. Template: http:
www.ieee.org/conferences_events/conferences/publishing/templates.html.
. Plagiarism: You cannot copy any sentences, paragraphs, or figures, from any external
sources (such as published papers or Internet). If Turnitin indicates that a submission is over
30% similar to any other submissions, the instructor will ca
y out a Plagiarism investigation.
c. If you have to cite a figure/graph published somewhere else, please properly cite the source of
the reference [0 credit if plagiarism check returns over 50% similarity to any published work].
[Grading of grammars and typos are included in the “Overall”]
2. Title and Abstract [1 pt]: Your report should have a
ief and informative title and an abstract. The
abstract should have XXXXXXXXXXwords, which summarizes the problem you intend to address in the
eport (e.g., text classification, transfer learning using naïve bayes classification etc.). Briefly describe
designs and solutions which will be proposed in the report, and
iefly summarize any conclusions
the report intends to draw. [ XXXXXXXXXXwords]
3. Introduction [2 pts]: Your report should have an introduction section with 500 – 1000 words. The
introduction should clearly state (1) what is the research problem to be studied in the report; (2) the
motivation of the problem studied in your report; (3) how are the problem solved by existing
methods, if any; and (4) a
ief description about the method you will propose in the report. You
should cite at least 8 relevant references (publications) in the introduction. [ XXXXXXXXXXwords]
4. Related Work [2 pts]: Your report should have a related work section to summarize works
(algorithms) which already exist to solve the problem. For example, if you are trying to address
transfer learning for naïve bayes classification using small number of training samples, you need to
discuss how transfer learning was used in naïve bayes classification, what are existing methods for
http:
www.ieee.org/conferences_events/conferences/publishing/templates.html
short message classification, what are common approach to transfer knowledge from a source, such as
email spam, to a target domain, i.e., the short message classification. etc. [ XXXXXXXXXXwords]
a. The related work section of your report must cite at least 10 peer reviewed references.
5. Main body [9 pts]: In the body of your project, you will need to provide technical details of your
design [ XXXXXXXXXXwords]
a. If you are proposing a new approach to transfer a naïve bayes classifier train from one dataset
to be used to another dataset, you will need to describe your designs. Describe the motivation
of your design, and why do you think this would work to solve the problems [2 pts]
. Use flowcharts, figures, or some pseudo-code to describe your algorithm details. [Please use
at least two figures (or flowcharts) to demonstrate the system framework or architecture] [5
pts]
c. If your report is about experimental studies, you will need to provide a
ief description about
your learning/classification methods, the benchmark datasets, and different measures applied.
You should also explain how the experiments are ca
ied out in your study, and what type of
empirical study goals you intend to achieve.
6. Experiments [6 pts]: In the experiments, you need to introduce (1) main purpose of the experimental
studies; (2) what are the tools used to design the algorithms; (3) what are the baseline methods for
comparisons; and (4) what are the performance measures and data used for empirical studies. You
should also use figures and tables to report the results collected from your studies, and summarize the
experimental results [ XXXXXXXXXXwords].
a. Experimental settings: including an introduction of baseline methods, programming
tools/languages, the setting of the parameters used for different methods. [1 pt]
. Benchmark data: Provide detailed description about data used for your study, including
detailed information about the size/dimension of the data.[1 pt]
c. Baseline methods: In order to demonstrate the performance of your method, you will need to
use a baseline approach, and compare the performance of your design with the baseline. For
example, a simple naïve bayes classifier trained from 5% of the target dataset (i.e., the SMS
dataset) can be used as a baseline to demonstrate that if a classifier is trained from a small
target set it might not perform very well. In this way, the experiments will demonstrate the
merits of your method, and validate your hypothesis [1 pt]
d. The results: The detailed results reported in figures/tables with necessary analysis and
descriptions. You will need to include at least one figure and one table to show the results. [2
pts]
e. Analysis of the Results: Please compare the performance of your method and the baseline
approach, and analyze why your method can obtain a good performance. Please also add a
case study example (e.g. an example of a review report and the predicted result from