Dear COMP9418 students,
We have released the exam marks, and they should appear in give in the next 24 hours.
Here is the exam marks histogram. The maximum exam mark is 100.
38.80 |2 |
45.60 |2 |
52.40 |2 |
59.20 |7 | ***
66.00 |6 | **
72.80 |11 | ****
79.60 |16 | *******
86.40 |11 | ****
93.20 |9 | ***
100.00 |2 |
-------- |------- | -------
N= |68
We also released the solution file for the exam. It is available on WebCMS and includes the test cases used for Q9 and Q10. Look at the exam solution carefully before reaching out to us.
Thank you very much for being a student of COMP9418. I (Gustavo) would like to thank the tutors (Yunrui and Zhijie) and course admin (Enver) for their assistance during the course.
Happy holidays!
Gustavo (on behalf of the COMP9418 team)
Dear students,
If you are still working on the exam, please remember that we are in the last 15 minutes.
Take your time to make a partial submission of your work.
Regards,
Gustavo.
Dear students,
Question 10 requires returning one elimination order with the expected width. Any elimination order with such width will be accepted as correct.
As in the tutorials, use a flat list to represent the elimination order.
Regards,
Gustavo.
Dear students,
The questions in the forum must not give away any piece of the solution. It has happened twice already.
Please, read the exam instructions about the questions.
Gustavo.
Dear students,
The COMP9418 exam will be online at 9 AM.
I will be online answering questions in the WebCMS forum and course email. Please allow some time for responses. I will take a few breaks during the exam for lunch, dinner and a couple of meetings.
Please, keep an eye on the WebCMS forum and your institutional email during the exam. Any updates to the exam description will be made via notices on WebCMS, which automatically send an email to all students.
You are allowed to make ten submissions of your work. The system is configured not to accept submissions after the deadline.
Good luck to everyone
Regards,
Gustavo.
Hi everyone, below are the final leaderboard results for assignment 2:
zid | average cost over 10 days | total runtime /s | bonus marks
z5410535/ 29135.6 315 10
z5310648/ 29165.0 54 9
z5427649/ 29226.4 19 8
z5473290/ 29509.0 230 7
z5441064/ 29590.2 89 6
z5208565/ 29890.2 319 5
z5060062/ 30001.4 1024 4
z5346386/ 30054.8 46 3
z5525268/ 30085.6 13 2
z5431400/ 30087.6 13 1
z5271686/ 30280.0 14
z5448686/ 31006.6 13
z5287859/ 31213.6 74
z5193638/ 31221.8 23
z5349951/ 31287.4 13
z5376977/ 31346.2 107
z5304797/ 31433.6 248
z5391857/ 32036.0 13
z5208477/ 32360.4 277
z5414409/ 33123.6 71
z5409374/ 33502.6 14
z5359871/ 33751.4 498
z5309909/ 34624.8 14
z5323706/ 34884.8 551
z5129534/ 35237.6 16
z5336891/ 35276.2 101
z5261614/ 35437.2 68
z3372932/ 36058.2 18
z5383398/ 36243.8 13
z5295964/ 36630.2 21
z5443808/ 37491.6 14
z5388136/ 37780.6 131
z5441033/ 43795.4 15
z5443505/ 47260.2 79
z5328465/ 47371.0 1289
z5352294/ 47670.0 14
z5431541/ 49341.4 92
z5519653/ 57013.8 357
z5335404/ 65423.8 13
z5315804/ 102620.2 58
z5409640/ 136486.8 26
Some submissions had errors in running: Please note that we cannot accept late file submissions. Feel free to email if you feel there has been another issue.
z5267156 | ModuleNotFoundError: No module named 'pgmpy'
z5279695 | ModuleNotFoundError: No module named 'BayesNet'
z5319088 | FileNotFoundError: [Errno 2] No such file or directory: 'HMMMatrix.json'
z5379229 | ModuleNotFoundError: No module named 'solution'
z5382403 | FileNotFoundError: [Errno 2] No such file or directory: 'data1.csv' (Further errors were found after replacing this file, please see assignment comments)
z5434537 | FileNotFoundError: [Errno 2] No such file or directory: 'data1_5_periods_early_morning.csv'
z5472298 | es = np.asarray([f[*(emission.get(x) for x in f.domain)] for f in self.e_factors]) SyntaxError: invalid syntax
z5282785 | FileNotFoundError: [Errno 2] No such file or directory: 'hmm_tables.pkl'
z3191524 | 31 minute runtime, too long to execute
z5375535 | 31 minute runtime, too long to execute
The final marks have also been released. Good luck with your exams everyone!
zID | Average cost over 10 days | Total runtime over 10 days (in seconds)
z5208565 | 29907.4 | 334
z5346386 | 30062.4 | 42
z5448686 | 30974.8 | 13
z5287859 | 31180.0 | 70
z5304797 | 31892.2 | 261
z5271686 | 31904.6 | 16
z5391857 | 32054.6 | 14
z5388136 | 35061.4 | 171
z5336891 | 35330.6 | 98
z5473290 | 35733.0 | 363
z3372932 | 36036.4 | 19
z5383398 | 36263.8 | 15
z5261614 | 37714.8 | 69
z5443505 | 47283.4 | 78
z5328465 | 47370.4 | 1757
z5319088 | 47576.6 | 114
z5376977 | 48000.0 | 11
5328465 | 48000.0 | 1860
z5441033 | 51840.6 | 47
z5414409 | 54204.6 | 57
3191524 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5060062 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5193638 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5208477 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5310648 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5375535 | Implementation runs too slowly (1860s)
5409640 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5410535 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5427649 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
5431400 | ModuleNotFoundError: No module named 'DiscreteFactors'
5525268 | ModuleNotFoundError: No module named 'solution' (please avoid submitting a .zip file and be careful that your file is named solution.py !)
Please submit your final submission correctly! Either using
give cs9418 ass2 solution.py report.pdf *.csv *.py *.pickle *.json
or selecting all relevant files individually when uploading to webcms. NO ZIP FILES!
Dear students,
We will run a second leaderboard today with all submissions. If you want to participate, ensure you submit your solution before 6 pm.
Regards,
Gustavo.
Hi everyone, below are the results for the leaderboard on Sunday.
ZID| Avg. cost | Time
z5525268 | 30040.4 | 14
z5193638 | 31298.0 | 22
z5208477 | 31666.6 | 271
z5271686 | 31891.6 | 15
z5391857 | 32045.4 | 13
z5336891 | 35199.4 | 96
z5388136 | 35363.6 | 178
z3372932 | 36019.8 | 18
z5383398 | 36179.4 | 15
Submission Errors:
3191524 | FileNotFoundError: [Errno 2] No such file or directory: 'data1.csv'
5060062 | ModuleNotFoundError: No module named 'BayesNet'
5208565 | File "/import/ravel/1/cs9418/23T3.work/ass2/leaderboard1/5208565/discrete_factors.py", line 10, in <module>
from typing_extensions import Self
ImportError: cannot import name 'Self' from 'typing_extensions' (/usr/lib/python3/dist-packages/typing_extensions.py)
5287859 | ModuleNotFoundError: No module named 'Graph'
5304797 | ModuleNotFoundError: No module named 'solution'
5310648 | ModuleNotFoundError: No module named 'solution'
5319088 | ModuleNotFoundError: No module named 'solution'
5410535 | FileNotFoundError: [Errno 2] No such file or directory: 'doorFactor.pickle'
5414409 | FileNotFoundError: [Errno 2] No such file or directory: 'data1.csv'
5427649 | ModuleNotFoundError: No module named 'solution'
5431400 | ModuleNotFoundError: No module named 'solution'
5443505 | ModuleNotFoundError: No module named 'solution'
5448686 | ModuleNotFoundError: No module named 'solution'
5473290 | File "/import/ravel/1/cs9418/23T3.work/ass2/leaderboard1/5473290/DiscreteFactors.py", line 115, in evidence
index = f.outcomeSpace[var].index(value)
ValueError: sequence.index(x): x not in sequence
Please make sure when you submit, you include all relevant files your implementation uses. Also, please ensure that you submit your solution exactly as "solution.py". I have modified the give submission to only accept this filename. We won't be accepting .zip files.
Due to filename issues and popular demand, we will be running another leaderboard due Thursday 6PM (16/11).
Dear students,
Welcome to Week 10.
In this final week, you will find lectures about learning structure and parameters in graphical models. The tutorials are about belief propagation and sampling.
Reminders:
I have made some new test cases for Assignment 1, Task 9. The old test cases computed the log probabilities for different class attributes using a structure created for class 'BC'. Although technically not incorrect, these test cases are strange because Naive Bayes is a classification model. The new test cases ensure the structure and the prediction match the same class variable.
Please run the test cases below if you did not get full marks for Task 9. If you receive higher marks with the new test cases, email me with the subject "Assignment 1, Task 9", and let me know your zID(s).
Regards,
Gustavo (on behalf of the COMP9418 team).
######################
# Test code to copy #
######################
outcomeSpace = learn_outcome_space(data)
naiveG = learn_naive_bayes_structure(data, 'BC')
naive_model = NaiveBayes(naiveG, outcomeSpace=outcomeSpace)
naive_model.learn_parameters(data, alpha=1)
evidence = data.iloc[42].to_dict()
del evidence['BC']
test(naive_model.predict_log('BC', evidence) == 'No')
outcomeSpace = learn_outcome_space(data)
naiveG = learn_naive_bayes_structure(data, 'AD')
naive_model = NaiveBayes(naiveG, outcomeSpace=outcomeSpace)
naive_model.learn_parameters(data, alpha=1)
evidence = data.iloc[42].to_dict()
del evidence['AD']
test(naive_model.predict_log('AD', evidence) == 'No')
outcomeSpace = learn_outcome_space(data)
naiveG = learn_naive_bayes_structure(data, 'Size')
naive_model = NaiveBayes(naiveG, outcomeSpace=outcomeSpace)
naive_model.learn_parameters(data, alpha=1)
evidence = data.iloc[42].to_dict()
del evidence['Size']
test(naive_model.predict_log('Size', evidence) == '<1cm')
outcomeSpace = learn_outcome_space(data)
naiveG = learn_naive_bayes_structure(data, 'Margin')
naive_model = NaiveBayes(naiveG, outcomeSpace=outcomeSpace)
naive_model.learn_parameters(data, alpha=1)
evidence = data.iloc[42].to_dict()
del evidence['Margin']
test(naive_model.predict_log('Margin', evidence) == 'Well-defined')
outcomeSpace = learn_outcome_space(data)
naiveG = learn_naive_bayes_structure(data, 'BreastDensity')
naive_model = NaiveBayes(naiveG, outcomeSpace=outcomeSpace)
naive_model.learn_parameters(data, alpha=1)
evidence = data.iloc[42].to_dict()
del evidence['BreastDensity']
test(naive_model.predict_log('BreastDensity', evidence) == 'medium')
Dear students,
Sorry for the spam, but I am seeking a research assistant for a project with Neura. We are building a natural language processing tool using Langchan and OpenAI API. If you have some experience or are interested in learning, email me. The implementation should take a couple of months, and the workload is about 10 hours/week. We can pay as casual.
Regards,
Gustavo.
Dear students,
You can find the course material for this week on WebCMS. This week's topic is approximate inference, and we cover Belief Propagation and Sampling. The tutorials will cover Gaussian Models, including Kalman Filters.
The myExperience survey is open. Please take a moment to answer the survey form before 23/11. The feedback from previous students helped us to improve COMP9418 significantly. Your feedback is greatly appreciated.
We will run a test evaluation for your model and release the cost and time of your models afterwards. This is to help you confirm that your model has no significant errors and works with the evaluation system. Please submit your answers by this Sunday (6 pm).
Finally, remember to submit your answers for Quiz 7 before Sunday at 6 pm.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
We released the marks for assignment 1. You can see them by assessing:
https://cgi.cse.unsw.edu.au/~give/Student/sturec.php
Please look at the test cases available on WebCMS ( https://webcms3.cse.unsw.edu.au/COMP9418/23T3/reso... ). If you have any questions, please email the course admin.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
UNSW has informed us that the COMP9418 exam will run on Thursday, 7/12/2023 .
The exam will last 12 hours, starting 7/12/2023 at 9:00 am AEDT and finishing at 9:00 pm AEDT. We will answer questions during the entire exam period. Any issues should be posted in the WebCMS forum (general matters) or emailed to cs9418@cse.unsw.edu.au (sensitive ones).
The exam will be available in WebCMS, and it will be a Jupyter notebook. You must write all answers in the notebook and submit them by give (command line or WebCMS) before the deadline.
The exam will have three parts:
The exam will require around 3 hours. Therefore, you must reserve at least 3 hours to work on the exam that day. We will not require you to submit the answers in 3 hours. The only requirement is to submit your exam before the deadline (7/12/2023 at 21:00 AEDT). No late submissions will be accepted. If you do not submit your exam on time, we will consider that you did not take part in the exam. You will need to apply for a supplementary exam.
Some students may have another exam on the same day (morning or afternoon). In this case, you should reserve your free period for the COMP9418 exam.
You will need a computer with the same libraries used in tutorials to solve the exam programming questions. You can use your computer, Google Colab or the UNSW CSE VLAB computers using software such as TigerVNC. The exam will be an open-book exam. Therefore, you can access papers, books, and course materials, including slides and solved tutorials. You cannot communicate (email, phone, message, talk, etc.) except with the COMP9418 staff via email or forum. Deliberate violation of exam conditions will be referred to Student Integrity as serious misconduct.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
UNSW has indicated that the COMP9418 exam date will be on the 7th of December. We have updated the exam page on WebCMS and the exams of the last three years, and corresponding solutions are available there. I will send a separate message regarding the exam.
The assignment 2 description is out. We will run a test evaluation on the 12th of November (Sunday). To participate in this evaluation, submit your solution before 12/November at 6 pm. This test evaluation is helpful to test your code against a large set of hidden cases. We will also distribute bonus marks to the top 10 submissions. Assignment 2 final submissions are due on the 19th of November.
Remember to submit your answers for Quiz 6 before Sunday at 6 pm.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
You can find the course material for this week on WebCMS. Do not forget to submit your answers for Quiz 5 before Sunday at 6 pm.
Assignment 2 description, data and sample source code are available on WebCMS. This assignment is due on Week 10, but we will run a leaderboard on Week 9 and give bonus points for the top-ten submissions. More details are in the assignment description.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
Happy flexibility week! This week we do not have any new videos, lectures, tutorials, quizzes or consultation hours. You can use this time to catch up with the course content.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
This week, we will cover undirected graphical models and MAP queries.
Here are a few important things you need to know about this week:
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
Enver will replace me in Week 5 lectures on 09/10 (Monday) and 12/10 (Thursday).
Monday's lecture on 09/10 (12-1 p.m.) will be delivered online through MS Teams. Use this link .
Thursday's lecture on 12/10 (4-5 p.m.) will be delivered in person at the Science Theatre (K-F13-G09).
Regards,
Gustavo.
Dear students,
I hope you had a lovely weekend and Monday holiday!
You will find on WebCMS the course content for this week. Do not forget to submit your answers to this week's quiz (due Sunday at 6 pm).
If your tutorial is on Monday, Zhijie has recorded his tutorial, and it is available here .
You are also welcome to join any other tutorial, including Zhijie's tutorial on Friday 12-2pm, if you prefer.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
Welcome to Week 3. This week, you will find on WebCMS:
Assignment 1 is available on WebCMS and is due 15th of October (Sunday) at 6:00 pm. You can work on this assignment individually or in a group of two students.
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
A friendly reminder that Quiz 1 is due this Sunday at 6 pm. We do not allow late submissions for quizzes.
Regards.
Dear students,
We have finished uploading the course material for this week. You will find on WebCMS:
Starting this week, we have a weekly quiz. The quizzes are multiple-choice questions. Please submit your answers before Sunday at 6 pm, and we will not accept late submissions for quizzes.
You can find the quizzes under the menu Activities in WebCMS.
The live lecture recordings are in Echo360. We have posted a link on WebCMS.
I am overseas this week. Enver agreed to replace me in the lectures.
Monday's lecture on 18/09 (12-1 p.m.) will be delivered online through MS Teams. Use this link .
Thursday's lecture on 21/09 (4-5 p.m.) will be delivered in person at the Science Theatre (K-F13-G09).
Regards,
Gustavo (on behalf of the COMP9418 team).
Dear students,
I will be overseas in Week 2 (18/09 to 22/09).
Enver (COMP9418 course admin) agreed to replace me in the lectures. Thanks, Enver!
Monday's lecture on 18/09 (12-1 p.m.) will be delivered online through MS Teams. Use this link .
Thursday's lecture on 21/09 (4-5 p.m.) will be delivered in person at the Science Theatre (K-F13-G09).
Regards,
Gustavo.
Dear students,
Some students have reported having issues accessing the course outline in ECOS. I have posted a PDF version here: https://www.cse.unsw.edu.au/~cs9418/outline_pdf.html
Regards,
Gustavo.
Dear students,
Welcome to COMP9418!
COMP9418 will run in the format of flipped classrooms. In the lectures, we will review the previous week's content on Monday and discuss one exam exercise in Thursday's classes.
One exception is Week 1, as we do not have content to review. I will present the course today and explain the course logistics. Thursday, we will solve a textbook exercise together.
You can find on WebCMS COMP9418 page (https://webcms3.cse.unsw.edu.au/COMP9418/23T3/) the following:
The material is online for all ten weeks, so you can study at your own pace.
The practical part of the tutorials is a Jupyter Notebook that runs on Google Colab.
Starting Week 2, we will have weekly quizzes, and the quizzes and other assessments are due on Sundays at 6 pm. Also, beginning in Week 2, we will post the previous week's tutorial answers on Monday mornings.
Therefore, check for new material on Mondays and submit your work before 6 pm on Sundays.
Today, we will have a shorter lecture at noon, and I will explain the course and answer your questions about its organisation. Starting next week, Monday's class will have a summary of the previous week's content + consultation. Thursday's lectures will discuss exam exercises.
I hope you will enjoy the course!
Regards,
Gustavo (on behalf of the COMP9418 team).