COMP7015 Artificial Intelligence Semester 1, 2023–24
Course Project Instructions
Instructor: Dr. Kejing Yin Oct. 1, 2023
1 Requirements
1. You should form a group of one to four people to carry out the project.
2. Milestones (important dates):
(a) Oct. 20, 2023 (Friday): Deadline of group information registration (names and student
IDs of your group member) and topic selection. You will be notified your group number
afterwards. Please register as early as you can to allow more time to work on the projects.
Link: https://hkbu.questionpro.com/comp7015-2324-reg
(b) Nov. 3, 2023 (Friday): Submission of a Progress Report in Moodle.
(c) Nov. 23, 2023 (Thursday): Submission of the final work, including all source codes, a
Final Project Report and a Presentation Video in Moodle.
3. In the Progress Report, you should briefly describe your motivation of the topic selection, work
progress (with contribution of each group member clearly stated), challenges encountered, and
plans to tackle the challenges in no longer than two A4 pages.
4. The Final Project Report should be at least five A4 pages, covering a brief introduction of the
topic you selected, an overview description of your methods used, detailed descriptions of the
experimental design, the results you obtained, a conclusion, and the contributions of each group
member. There is no upper page limit, but it is suggested not to exceed eight A4 pages.
5. In the Final Project Report, you are required to faithfully report the contribution of each group
member. Your final score will depend on how much contribution you make to the project.
6. The Presentation Video should be 10-15 minutes, covering your solutions to and the main results
of the course project. Each group member should clearly present his/her contribution to the project
and the entire presentation should be organized in a logic way.
7. In the Presentation Video, you are required to display the presenter’s faces clearly to allow the
lecturers to identify your identity.
8. You are highly encouraged to use as much visualizations as possible (e.g., figures, tables, and
graphs) to make your presentation clearer.
9. In the final submission, you are required to submit three files to Moodle: 1) the Final Project Re-
port in .pdf format, 2) the presentation video in any video format, and 3) all other files/documents
and the source codes compressed in a .zip file.
10. We have zero tolerance to cheating and plagiarism. Your project report and source codes will be
scanned by anti-plagiarism platforms, anyone who plagiarizes and whom were plagiarized will
receive zero score and be reported to the departmental exam committee for further penalty.
11. Resubmission of previous work: Please be noted that you cannot submit your previous work
(e.g., published papers and work submitted to other courses) or otherwise it will be considered
self-plagiarism.
12. Third-party library policy: In option 2 (adversarial search), you need to implement the algo-
rithms on your own. Using third-party libraries are not allowed. In options 1 and 3, you can use
third-party libraries but you need to explicitly acknowledge them in your report and presentation.
1
Course Project Instructions 2
13. If you take a cue from other sources, e.g., books, literature, and online resources, you should
clearly cite them in your Final Project report. Note that it is explicitly prohibited to copy source
codes from online resources or that generated by generative AI tools like ChatGPT.
2 Topics
You can choose one topic from the following:
Topic 1: Coupon Acceptance Prediction
Suppose you are a data scientist working at a company that develops a mobile payment app. You are
responsible for making recommendations of different coupons to vehicle drivers. To make more effective
recommendations, you would like to build a machine learning model to predict whether the driver would
accept the coupon or not.
For this problem, we can use a dataset collected via a survey on Amazon Mechanical Turk. The
survey describes different driving scenarios including the destination, current time, weather, passenger,
etc. A detailed description can be found in https://archive.ics.uci.edu/dataset/603/
in+vehicle+coupon+recommendation.
In this project, you are required to:
1. Select 3–4 suitable machine learning models for this dataset. Introduce the models in your reports
and presentation.
2. Select proper method (hold-out, K-fold cross validation, etc.) to evaluate the performance.
3. Measure the performance using the metrics introduced in the course. Compare the different ma-
chine learning models using the metrics. Discuss the differences between different performance
metrics.
4. Use appropriate regularization techniques and compare the performance before and after applying
regularization.
5. Try to turn the hyperparameters in the models you selected and observe how they affect the per-
formance.
Topic 2: Adversarial Search for Tic-Tac-Toe
In this topic, you are required to implement the adversarial search algorithms, including minimax and
alpha-beta pruning algorithms, to develop an agent that plays the tic-tac-toe game. In this game, two
players will take turns to mark the spaces in a 3 × 3 grid with X or O. The player who succeeds in
placing three of their marks in a horizontal, vertical, or diagonal row is the winner.
In this project, you are required to:
1. Implement the functions necessary for formulating this deterministic game.
2. The game developed should be able to accept input from human player (either terminal or GUI is
acceptable) and take actions that maximize the agent’s score.
3. Implement the minimax search algorithm to determine the next move.
4. Implement the alpha-beta pruning algorithm and analyze the improvement in terms of running
time.
Course Project Instructions 3
Topic 3: Open Topic
If you are not interested in the two problems introduced above, you can propose another problem/dataset
to work on for your course project. If you decide to do so, there are a few more requirements:
1. Submit a project proposal (at most two A4 pages) by Oct. 20, 2023. You should clearly describe
the problem you would like to work on, the motivation of choosing the topic, and your plans to
carry out the project. You should determine the topic when you form the team.
2. Since there is only about 6 weeks for the project, so the scope should not be very large. Having
said so, the workload should be at least similar to the two problems above.
3. In the project proposal, you should briefly state the problem and precisely describe what you plan
to do, including what dataset or algorithms you plan to use, how would you evaluate the results,
etc.
4. You do not need to limit yourself to the algorithms/models introduced in the course. Any AI-
related methods can be used as long as they are suitable for the problems you proposed.
5. If needed, the course instructor and TAs will give brief feedback to your proposal to make sure
that it is feasible.
Appendix: Course Project Assessment Rubrics
Overview
The assessment of the course project is based on the implementation, the final report, and the
presentation video that each group submits. Each group member will be given an individual
score based on his/her contribution to the project and the quality of presentation. The weighting
of each item is as follows.
1. Implementation 30%
2. Final Report 35%
3. Presentation 35%
Rubrics for the Final Report and Presentation
Introduction to
the topic selected
and descriptions
of the methods
used (5)
The topic selected is
introduced, and the
methods used are described
in detail. The reasons of
using the algorithms are
explained.
(4-5 marks)
The methods used are
described but contains
some mistakes. The
reasons of using the
algorithms are less
convincing.
(1-3 marks)
Does not introduce
the methods used.
(0 mark)
Detailed
descriptions of
the experimental
design / proposed
solution (10)
The question to be
answered is clearly
defined, and experimental
design or the proposed
solution (for option 3) are
described in detail.
(7-10 marks)
The experimental design
is described but contains
some mistakes.
(1-6 marks)
Does not introduce
the experimental
design.
(0 mark)
Results Analysis
(12)
The results obtained are
clearly presented using text
and proper visualizations,
correct conclusions are
drawn from the results
obtained.
(8-12 marks)
The results and
conclusions are
presented, but they
contain some mistakes.
(3-7 marks)
Does not present the
results, or the results
are not organized/
explained and cannot
be read and
understood.
(0-2 mark)
Contributions of
each group
member (3)
Detailed contribution of
each group member is
clearly described.
(3 marks)
Contribution of each
group member is
mentioned but not in
detail.
(1 mark)
Does not mention the
contribution of each
member.
(0 mark)
Clarity,
organization, and
Logic (5)
The report and presentation
are well organized. The
logic is clear and easy to
follow. (4-5 marks)
The logic and
organization are
understandable but
could be further
improved.
(2-3 marks)
The logic and
organization are
confusing. The report
and presentation are
hard to follow.
(0-1 marks)
Course Project Instructions 4
Rubrics for Implementation
(1) Topic 1: Coupon Acceptance Prediction
Choosing
Machine
Learning
Models (5)
Proper machine learning
algorithms are chosen
for the task with
convincing reasons.
(4-5 marks)
The model chosen are okey,
or the reasons for choosing
the algorithms are not
convincing.
(1-3 marks)
The models chosen
are wrong for the
defined task.
(0 mark)
Model Training
(5)
The machine learning
models are properly
trained.
(4-5 marks)
The models are trained but
the implementation contains
mistakes.
(1-3 marks)
Does not train the
models or the
implementation is
wrong.
(0 mark)
Model
Evaluation (10)
Correct strategies for
evaluation and proper
evaluation metrics are
used and correctly
implemented.
(7-10 marks)
The evaluation strategies and
metrics are mostly correct,
but the implementation
contains mistakes.
(1-6 marks)
Does not evaluate
the model
performance.
(0 mark)
Hyperparameter
Tuning (5)
Correctly identify the
hyperparameters and do
hyperparameter tuning.
(4-5 marks)
Hyperparameter tuning is
done but the implementation
contains some mistakes.
(1-3 marks)
Does not attempt to
do hyperparameter
tuning.
(0 mark)
Efficiency and
readability (5)
The source codes are
easy to read, and some
efforts are made to
improve the efficiency
of the algorithm.
(4-5 marks)
The source codes are
understandable and can be
executed, but it could be
further improved.
(1-3 mark)
The source codes
cannot be
understood or
cannot be executed.
(0 mark)
(2) Topic 2: Adversarial Search for Tic-Tac-Toe
Implementing
the game setup
(5)
The game is correctly
implemented and
accepts input from
human player.
(4-5 marks)
The implementation
is okey, but there are
some errors.
(1-3 marks)
The game environment is
not correctly implemented,
or it does not accept input.
(0 mark)
Implementing
the functions
formulating the
search problem
(5)
The necessary functions
are implemented and
error-free.
(4-5 marks)
The functions are
implemented, but the
implementation
contains mistakes.
(1-3 marks)
Does not implement the
necessary functions or the
implementation is wrong.
(0 mark)
Implementing
the minimax
search (10)
The minimax search
algorithm is correctly
implemented.
(7-10 marks)
The algorithm is
implemented but
contains error.
(1-6 marks)
Does not implement the
search algorithms.
(0 mark)
Implementing
the alpha-beta
pruning (6)
Correctly implements
alpha-beta pruning
algorithm.
(4-6 marks)
The implementation
contains some
mistakes.
(1-3 marks)
Does not implement the
pruning method.
(0 mark)
Analysis of the
alpha-beta
pruning (4)
Proper profiling is done
to analyze the advantage
of alpha-beta pruning.
(3-4 marks)
Profiling is done but
there are some errors.
(1-2 mark)
Analysis of the alpha-beta
pruning is not carried out.
(0 mark)
Course Project Instructions 5
(3) Topic 3: Open Topic
Choosing AI
Algorithms for the
Proposed Problem
(5)
Proper AI algorithms
are chosen for the
proposed problem with
convincing reasons.
(4-5 marks)
The algorithms chosen
are okey, or the
reasons for choosing
the algorithms are not
convincing.
(1-3 marks)
The models chosen are
wrong for the defined
task.
(0 mark)
Algorithm
Implementation (10)
The algorithm chosen
are properly
implemented.
(7-10 marks)
The algorithms are
implemented but the
implementation
contains mistakes.
(1-6 marks)
Does not implement
the algorithms chosen.
(0 mark)
Algorithm
Evaluation (10)
Correct strategies for
evaluation and proper
evaluation metrics are
used and correctly
implemented.
(7-10 marks)
The evaluation
strategies and metrics
are mostly correct, but
the implementation
contains mistakes.
(1-6 marks)
Does not evaluate the
model performance.
(0 mark)
Efficiency and
readability (5)
The source codes are
easy to read, and some
efforts are made to
improve the efficiency
of the algorithm.
(4-5 marks)
The source codes are
understandable and
can be executed, but it
could be further
improved.
(1-3 mark)
The source codes
cannot be understood
or cannot be executed.
(0 mark)
Course Project Instructions 6