UNSW Course Reviews

[Opengangs]

Subject Code/Name: COMP4141 - Theory of Computation

Contact Hours:
- 2 x 2 hour lectures
- 2 hour tutorial

Assumed Knowledge:
Undergraduate: The formal prerequisites are: MATH1081, and (COMP1927 or COMP2521).
Postgraduate: The formal prerequisites are: COMP9020, and COMP9024.

Assessment:
- Assignments (50%; 12.5% each)
- Final exam (50%)

Lecture Recordings? Yes.

Notes/Materials Available: Lecture slides and tutorial solutions are available.

Textbook:
Recommended textbook: Introduction to the Theory of Computation by M. Sipser.

Lecturer(s): Dr. Paul Hunter

Year & Trimester of completion: 2022 Term 1

Difficulty: 4/5

Overall Rating: 4/5

Your Mark/Grade: 90 HD.

Comments: The rating would be a 5/5 if the issue of administration were resolved but alas, even after a year, it has not changed. The course material is really fascinating but definitely not a typical COMP course. If you decide to take the course, be prepared to work your butt off. The assignments are really fun to do, but they can take so much of your time if you're not up to date with the course material.

The main complaint of the course (and the sole reason why this is deserving of a 4/5 and not a 5/5) is because of administration. It seems somewhat ridiculous to receive marks or feedback (if at all) so late in the term, especially when all of the assignments have been submitted. It makes the feedback that we receive meaningless because we can't use it to improve on the quality of the work. If feedback and/or marks were returned back to us in a timely fashion, then I may consider bumping this up to a 5/5.

[RuiAce]

Subject Code/Name: COMP9417 - Machine Learning and Data Mining
Equivalent postgraduate: COMP9417 (identical course code)

Contact Hours: 2x2hr lecture, 1hr tutorial

Assumed Knowledge:
Undergrad: Two pathways: (MATH1081 + either COMP1531/COMP2041), or COMP2521
Postgrad: COMP9020 + COMP9024

Data structures and algorithms (both UG and PG) and knowledge of python suffices for the computing aspect. But you really should know some calculus, linear algebra, and statistics, in preparation for the math side.

Assessment:
- 1 x 1% homework
- 2 x 7% homeworks
- Weekly questions from tutorial set, best 7 out of 8 counts, 5%
- 30% project - hackathon, or comp9417 group project
- 50% 90min final exam

Lecture Recordings? Yes, on Microsoft Teams and UNSW Echo

Notes/Materials Available: Relatively detailed lecture slides and tutorial sets. Half of the labs were very in detail; presumably all labs will be in detail next term. Also some supplementary youtube recordings from the head tutor. Head tutor managed the course forum very actively. Overall surprisingly abundant set of course resources. (However, the internet is still a valuable resource for more niche concepts.)

Textbook: No single textbook recommended anymore. A list of optional textbooks for further reading provided on the course outline on webcms3, but I didn't use any of them.

Lecturer(s): Dr. Gelareh Mohammadi

Year & Trimester of completion: 22T1

Difficulty: 3/5 (however hackathon can boost this up to 4.5/5)

Overall Rating: 4/5

Your Mark/Grade: 97 HD

Comments:
This is one of many Ai courses offered at UNSW. At this point I really feel "machine learning" is a buzzword, but the course outline definition is loosely speaking enough. Namely that ML is the algorithmic approach to learning from data. It can be perceived to have a similar goal to statistical modelling, but in ML prediction accuracy tends to overrule interpretability of the model.

The course introduces some classical ML techniques, but also touches on pieces of the current state-of-the-art models (e.g. ensemble learning, neural nets). There's quite a lot of content, but this is to be expected since ML is currently rapidly growing. Generally speaking it is a good overview to current ML techniques though. (Surprisingly, it's also made me appreciate neural nets more, despite only spending 1 week on it.) As a result of so much content though, the lectures were quite fast paced. For a math major like me i didn't care, but I can see it being difficult for other students.

I should direct your attention to this review briefly, and how the final exam dragged a 3/5 down to a -5/5. Thankfully that was over. No idea if the different lecturer meant anything here, but my exam was essentially 50 MC. Not a great experience per se - the curveball questions were quite hard. But the exam didn't feel evil or bizarre at all.

What hurt the rating? Well, homework 0 was a grind for just 1%. Not a hard to get 1%, but tiresome. As the course progressed, this was kind of forgotten, because both subsequent homeworks were interesting and made up for it. Then it came to the project. In all fairness, the hackathon itself was interesting - good final goal we were aiming for, and it gave a taster of real world data science. Have to go self learn stuff (e.g. modelling beyond 9417 scope, mastery of pandas), but that's okay. What sucked was the server and the restraints. Painfully slow to work on AWS against 50 or so groups, all trying to fit these high CPU consuming models all the time. Could only make 10 submissions a day (previously 1 submission a day which was worse), so hard to get better model performance. Some stuff just took hours to run. (Also personally salty at the final rankings.) Difficult to tune models as well. Overall killed the course experience. Nevertheless, didn't throw the course down in the dumps or anything.

Also take some caution with the last topic on learning theory. Very interesting, but much more mathematical. Might be hard if you haven't done theoretical comp sci. But it only lasts a week, so it can't torture you too much if you hate it.

[anomalous]

Subject Code/Name: MATH3371 - Numerical Linear Algebra

Contact Hours: 3x 1 hour lecture, 1x 1 hour tutorial, 1x 1 hour lab class

Assumed Knowledge: One of the following:
- MATH2501/2601
- MATH2019 with at least a DN
- MATH2099 with CR

Basically, any course beyond 1st year which covers linear algebra. Knowing some programming is an implicit requirement here too: you can use either MATLAB, Julia or Python for the course.

Assessment:
- 3x quizzes, worth 5% each and delivered via Mobius (not as bad as you think, but still, ugh)
- Class test, worth 20%
- Assignment, worth 15%
- Final exam, worth the remaining 50%

Lecture Recordings? Yes, on Blackboard Collaborate.

Notes/Materials Available: Full detail course notes are written, but the slides are also provided. The notes are a bit more terse than the slides, so for some topics I found it was actually nicer to read the slides instead.

Textbook: No textbook required, and none recommended either.

Lecturer(s): AProf. William McLean, Dr. Quoc Le Gia

Year & Trimester of completion: 22T1

Difficulty: 3.5/5

Overall Rating: 4/5

Your Mark/Grade: 88 HD

Comments:
This is a brand new course, so no one really knew what to expect, but I decided to take a chance on it since it seemed like an interesting mix of maths and computing. Having done so, I feel this was a decent pick. The next run in 2023 will probably feel a bit better to those taking it, since the course content was still sort of in development during the term, and they’ll have feedback from our cohort to use for further changes. To be objective though, the course was a little bit unpolished this term.

The focus of this course is, naturally, numerical linear algebra. Linear algebra is everywhere, and often the method you learn in the theoretical linear algebra courses for doing various things (e.g. finding the eigenvalues of a matrix) is too slow, totally impractical or prone to precision errors in the real world. As well as seeing the derivation/justification for these more practical methods of performing linear algebra, you also analyse how their computational costs scale and, where possible, how different methods compare. While the material in the first half of the course is pretty chill, things start picking up after flex week: the numerical analysis in the accuracy and reliability topic is no joke, and conjugate gradient methods can be a bit tough to get your head around at first. Probably the most confusing part of this course though is the sudden inclusion of a small “machine learning” topic on SVMs at the end, which felt very out of place due to its notable lack of any actual linear algebra (instead, it was basically all nonlinear optimisation).

Given that the course is inherently computational in nature, part of the course work involves programming. The lab tasks and assignment are intended to reinforce this in a hands-on way, while the rest of the course dealt with the theory aspects. The labs weren’t assessed directly, and I didn’t really see the point of them until the final exam, because you were basically required to use numerical computation to answer one question (which made me regret not doing all of the labs; I gave up in probably week 4).

[fun_jirachi]

Subject Code/Name: MATH3161
 - Optimisation

Contact Hours:
2x 2 hr lecture
1x 1 hr tutorial

Assumed Knowledge:
Prerequisite: 12 units of credit in Level 2 Mathematics courses including MATH2011 or MATH2111 or MATH2510, and MATH2501 or MATH2601, or both MATH2019(DN) and MATH2089, or both MATH2069(CR) and MATH2099.

Assessment:
2x Class Tests (one weighted 15%, one weighted 20%)
1x Assignment 5%
Final Exam 60%

Lecture Recordings?
Yes

Notes/Materials Available:
Yes - all course provided (Moodle or otherwise)

Textbook:
None

Lecturer(s):
Professor Jeya Jeyakumar

Year & Trimester of completion:
22T1

Difficulty:
2.5/5

Overall Rating:
4/5

Your Mark/Grade:
87 HD

Comments:
Not a hard course but definitely a course where you have to keep yourself motivated because there is a LOT of raw computation to get through. There are certain things that the course assesses that are painful to work through, namely line searches and Hamiltonians, but this view is admittedly subjectiveness because of my comparative unwillingness to do continuously practice the computation required to solve questions of this sort. My only other gripe with the course was the assignment, which seemed unnecessarily verbose and long for what the outcome turned out to be, but even then this clutches at straws.

Otherwise, the course was excellent without being brilliant; I got more or less everything I wanted out of the course without it being as impressionable as some of the previous courses. This course has been and remains consistently very good, any maths student that has time to take this course should do at some point.

[yliu5532]

Subject Code/Name: MATH2111 - Higher Several Variable Calculus

Contact Hours: 
2*2h lectures, 1*1h lectures, 1h tutorials

Assumed Knowledge:
a mark of 70 or above in MATH1231/1241/1251

Assessment: 
Mobius quiz + written proof *2 (10% + 20%, in week 3 and week 7 respectively)
70min Class test (20%, week 10)
3h Final exam which tests the whole course (50%)

Lecture Recordings?  Yes

Notes/Materials Available:  Lecture notes will be made available on moodle

Textbook: N/A

Lecturer(s):
Dr Anita Liebenau - 4.9/5. She is the LIC of the first component (multivariable calculus and real analysis) which falls under pure mathematics. She explains everything very clearly, but the lectures can get a bit dry. It might be due to the content that at times her lectures isn't super engaging. Also, her lecture slides are hand written so it might be hard to find a keyword compared to the typed slides in MATH1251. Overall great lecturer, she made me understood most the concepts, and she is so responsible that she always ask students to type questions into the zoom chat. One thing to note, she is really nice when it comes to marking proof questions
Apro Guoyin Li - 5/5. I'm happy to give more if theres an option for it, but yeah, he is the absolute beast here. He's so passionate towards applied mathematics that he sometimes laughs while teaching. He is the LIC for the vector calculus component of the course, which falls under applied maths. He explained the notes with great detail, and, although vector calculus can be very computational, he tried to explain all the relevant theory behind each formula that we use. As a result, I understood everything that he taught and smashed that class test + the second part of the finals.

Year & Trimester of completion: 22T1

Difficulty: 2/5

Overall Rating:  4.9/5 (I would give a 5 if Anita's lecture slides were typed, but this might just be me who kinda not liked the way it was)

Your Mark/Grade: 80 DN

Comments:
The course is structured quite nicely overall. The first two weeks can be very dull as it is all about those confusing limits stuff (open and closed sets, path-connectness and stuff), while the remaining weeks of content are very interesting. Differentiation could be a bit dull but integration and fourier series are really fun, the later stuff like path integrals and line integrals and their applications were a breeze, and integral theorems are pretty straight forward as long you kept up to date with everything. I kinda left everything at the last minute and didn't really had much time to do practice problems + I didn't manage to figure out the first proof question which was worth 1.5% of the course. With all those issues together, I still managed to scrape an 80, so it can be a wam booster if u actually do your tut questions and past papers (like how u study for the hsc). The main issue for me in the finals was time management since I wasn't too familiar with every concept i had to look back to the lecture slides a lot, but with that being said, the final exam is significantly easier than the first year 1151/1251 exams conducted on mobius, so if u are choosing the course, stay up to date to ensure that its a wam booster. Lastly, i strongly recommend this course (either over MATH2011 if u met the mark cutoff or as an elective if ur a compsci or science student), the course content and lecturers are amazing and 100% worth doing! It is probably the best maths course ive taken so far (compared to this 2621 was absolutely painful).