COMP364 Final Project

Introduction

The bulk of class time for the final two weeks of the semester will involve work on the final project. There will be no other homework assignments during this time. Students will present the results of their projects in the final exam slot. Thus, the project should represent three weeks of work for this course, or (very roughly) 25-30 hours of work.

Grading

• Code (45 points): The project must involve a substantial programming component. Any programming language may be used. The programming may build on an existing code base from a previous assignment or downloaded from any suitable source. Code written for the final project should be clearly marked using comments that contain the string "final project". Although the amount of code written will vary considerably between projects, this part of the project should represent 10-15 hours' work, which would typically translate into a few hundred lines of code. Code will be graded on the same criteria as the programming assignments in this course, as listed on the course web pages.
• Written report (45 points): The objectives, methods, and results of the project should be described in a formal written report. The overall style of the report should be that of a scientific publication, such as the ones we have studied in this course. The audience of the report should be the instructor and other students in the course. Thus, notions covered in the course may be assumed without definition, but other concepts should be carefully described in the report. As with any college-level writing, suitable citations should be used wherever appropriate, and a complete bibliography must be included. Any suitable structure for the document may be used. One possible structure is to use the following section headings:
  • Introduction: describe the goals your project and any context or background required.
  • Methods: describe the design of your code and your experiments.
  • Results: describes the results of your experiments, including tables and charts if appropriate.
  • Conclusions: summarize what you learned from the experiments; optionally, discuss alternative approaches and further work that could be done.
  • Bibliography: list the sources used.
  Any reasonable spacing and formatting conventions may be used. The suggested length of the report is 5-10 pages, but longer reports will not be penalized. The report will be graded on the usual characteristics for writing (clarity, correctness of spelling and grammar, completeness in addressing the above requirements, logical structure), together with the overall intellectual merit of the project.
• Presentation (10 points): In the final exam slot, the project and its results must be described in a presentation of approximately 10 minutes' duration (presentations shorter than 7 minutes or longer than 15 minutes will be penalized). A rubric for this presentation is provided; it is very similar to the rubric for the paper presentation earlier in the semester.

Milestones

The project consists of three milestones: FP1, FP2, and FP3. FP1 and FP2 are ungraded, but are designed to assist with keeping the project on track. FP3 consists of the submission of code, report, and giving the presentation in the final exam slot. Specifics of the milestones are as follows:

• FP1: Submit to Moodle a 100-300 word paragraph describing:
  • The topic of your project.
  • A minimal coding milestone that you are absolutely sure will be achievable within one week of work, and will permit some interesting experiments to be conducted and written up. (Obviously, it would be preferable to achieve more than your minimal milestone, but the objective here is to ensure you have something to write about in your report, since you will need to start writing the report well in advance of the final deadline.)
  • A brief description of the proposed experiments that will be written up.
• FP2: Submit to Moodle a zip file of the code written so far, and a brief written update describing your progress relative to the minimal milestone identified in FP1.
• FP3: Submit to Moodle a zip file of all the code you wrote, and the written report.

Potential project topics

Here is a list of potential projects:

• Implement A* search with a landmark-based heuristic, and compare performance with other search algorithms using real road data.
• Further improve your Othello player using one or more of the additional tricks mentioned in the textbook (quiescence, forward pruning, beam search, ProbCut, adding a database of openings and/or endgames).
• Implement expectiminimax for a simple stochastic game and investigate the performance of various evaluation functions, including one based on Monte Carlo simulation.
  • Note: For this project and the next one, here is one simple two-player stochastic game that might be suitable for analysis. The game, which we'll call Monopole, is played on a board with 20 locations in a circuit (think Monopoly). Both players begin on the designated start location. Every time you pass the start square, you receive $200 from the other player (similar to, but not the same as, Monopoly). At each turn, you may choose to remain where you are, or to move by rolling two 6-sided dice. If you choose to move, you roll the dice and move forward by the number of locations which is the sum of the two dice. However, if you end up on the same location as the other player, you must immediately pay that player $1000. The game ends after each player has had 10 turns.
• [Adapted from textbook exercise 5.17] Implement expectiminimax and *-alpha-beta (as described by Ballard (1983) -- see the textbook bibliography) for a simple stochastic game; compare their performance and analyze how well the pruning works.
• Implement simulated annealing and/or a genetic algorithm on one or more interesting problems (for example, image denoising, traveling salesman) and conduct experiments on their effectiveness. See textbook exercises 4.3 and 4.4 for more specific ideas.
• Tackle one of the projects suggested in Section 10 of the document describing the Clue programming assignment: counting cards, a DPLL implementation, or a stochastic local search implementation.
• Extend your decision tree implementation from programming assignment 4 to handle missing data and numeric data, and implement some version of pruning. Conduct experiments to determine whether pruning improves the performance of the tree and if so by how much.
• Implement the back-propagation learning algorithm for a simple neural network, and conduct experiments to determine its efficacy.

Final remarks