I think the winter 2024 offering of C147 was very similar to the reviews I've read in the past:

- There were some very hard/long assignments. However if you used Piazza you should be fine
- The class was a bit more mathy in the beginning than compared to the end
- The professor was excellent!

There are also several comments complaining about the "slow pace" of the class, "over-enrollment" and new 50% Midterm Weight. I'd like to address those and what my views are as a fellow student:

Personally, as someone who skipped M146 and went straight into C147, I appreciated that the professor reviewed some of the more fundamental aspects of DL in the beginning of the course. Furthermore, I'd like to point out that this is after all, a graduate class. Grad Students often don't get to take the particular sequence of courses at UCLA prior to coming here, so it make sense that the professor do some review in the very beginning to make sure everyone is on the same page.

For those that were interested in "practical" machine learning (ie learning one of the popular ML libs), the final project was an opportunity to do that. Personally, I think this class is not a torch-bootcamp and rightfully so: C147 focus on more fundamental knowledge and the theory side of DL. There are also other more coding-heavy classes offered (CS188 in W24 for example) that one can take to gain that kind of experience.

While it is true this was a very big class, personally between taking the class vs not taking the class I'd always choose the former. I definitely appreciated the effort the teaching staff put into making the logistics work for such a large class!

Finally, the median for the class was an A- (partly due to grad class, also because Kao doesn't grade harsh). I really don't understand why people complain about grades at this point considering this is an upper-div and the midterm median was a B.

I don't have anything to say that others haven't already said, Professor Kao is truly one of the best lecturers at UCLA and I would highly recommend this class if you are interested in Neural Networks and Deep Learning. Also, the TAs for this class were amazing, especially Tonmoy Monsoor. Tonmoy is insanely knowledgeable about the topic and his discussions were super useful for the homeworks!

Grading:
Homework: 40% (5 homeworks)
Midterm: 30%
Final Project: 30%
Extra Credit: 0.5% for filling out class eval, up to 1.5% for participating on piazza (in a useful way), and some extra credit given on the midterm (final question on the exam is optional extra credit)

I would highly recommend this class to any interested in deep learning and machine learning. Professor Kao is a very good lecturer and he does an amazing job explaining concepts. I never truly understood how backpropagation worked until he explained it in class. Anyone interested in research/ML should definitely take this class. You will learn so much.

However, the class is not a cake walk. It's actually fairly easy to get a good grade in this class as long as you put in the effort. There is only one exam around week 8, which won't be bad if you pay attention to lecture (our average for the exam was a 95%). The homeworks are the real killer and can take a very long time. You essentially have to build neural networks from scratch using Python and Numpy.

Overall, this is an amazing class where you can truly learn so much, but at the price of many hours of homework. Professor Kao is probably one of my favorite professors I have ever had at UCLA.

The first part of this review is to the people who are considering taking this class without prior experience. I highly recommend not taking this class unless you took M146 and have some experience in machine learning, both of which I didn't do (this my fault). The course was very math heavy at the start and Kao doesn't define many of the ML terms that he already expects you to know.

Generally, the workload is also very intense and very much requires that you have an understanding of numpy (it will be extremely painful if you do not). Luckily Tonmoy was very helpful in his office hours for the homeworks, but the homeworks will generally be awful.

The class is very theory based. You learn a lot of how neural networks functions and the function of each hyperparameter, but you won't be taught much of how to use frameworks such as PyTorch or good practices for training a model.

I also personally would have changed the grading scheme a bit. 50% for midterm is a bit excessive. There was also double jeopardy on the backprop question on the midterm: if you made the same small mistake on both sides of the backprop, you would lose the points for both sides; you could lose 4% of your total grade in the class just for making a small algebra mistake. The 2% extra credit is nice though, and the project is graded very leniently.

I don't know what more I can say that other reviews do not state already. Professor Kao is one of the most helpful and kindhearted professors I have had the utmost pleasure to learn from. His use of slides, recorded lectures, zoom livestream, have all helped me keep up with the class without having to worry about missing one or two lectures. The TAs are the best TAs I've ever encountered in my time at UCLA. Yang, Tonmoy, Kaifeng, Shreyas, and Lahari were very helpful; were straightforward with you if you got a question on the homework right or wrong (they don't dance around you and say 'hmm you might be right' or give you some BS answer), no, they help you get to the right answer if you're stuck and they corroborate you if you are correct. The midterm was hard, but expected. The questions mirrored the midterm review closely as the TAs had emphasized, and the TAs are straightforward with you if you ask a question about what's on the midterm. I asked one of them, 'is expectation going to be on the midterm,' to which they simply replied, 'yes.' Office hours were an absolute godsend. Go to them if you are not comfortable with the subject. I had satisfied absolutely NONE of the pre-reqs, so I went to OH to get the help I needed, and it WAS helpful.

I won't sugarcoat it; this class is A LOT of work. It's fairly easy to get an A, but be ready to also put in the time and effort to achieve that grade. I dedicated around 10-15 hours every week to this class (I took CM146, CS143, and DH101 as well for reference). It was highly rewarding and I learned SO much. AI was such a blackbox before I took this class; there was so much hype and pizzazz surrounding it. But after taking C147, it really broke it down into the base parts that go into building a neural network, and though I no longer look at AI mystically, I enjoy learning about it nonetheless. So, for anyone who is interested in this subject or is looking for a CS elective, take C147.

Kao teaches this well. I didn't have any of the prereqs and did fine. Just start assignments early and go to discussions. Many people did not study very much for the exam this year which is why it was lower than previous years. In my opinion we had by far the easiest exam (but the extra credit was very difficult) compared to previous years. The only prereq you really need is multivariable calculus, knowledge of what expectation is, and the most important is probably python and numpy skills. The rest will come. I wish we covered more material, lots of students asked really bad questions during class which kept us behind. Still recommend if you are very interested in deep learning. If you aren't very interested, you may not like the class.

Needless to say, Prof. Kao is an amazing lecturer. His teaching style is the best I have ever seen in the entirety of my undergrad career. But you already knew that. This is the third class I have taken with him, but unfortunately, it was the worst one. The class size (>500) made everything unbearable, as people were constantly asking questions. This demonstrably hindered lecture progress, as we ended up about 4 (?) lectures behind. These sometimes came in the "trying to appear smart in front of Professor Kao" flavor - thank you sweaty CS majors. Piazza, which is our class discussion forum, went from okay to terrible over the course of the quarter. Besides the fact that participation is hard to get because the average response time is under 5 minutes (thank you again to the literal hundreds of sweaty CS majors), people ended up mass posting random things effectively begging for participation points after the midterm. This midterm was more difficult than past years' exams, and it was changed from 30% of the grade to 50%. The midterm grades were noticeably lower, but not enough to warrant a curve which is completely fine. I just think the exam format should not be half of the grade, and it shouldn't be dependent on how much of the content from the TA's review sessions can you stuff onto your cheat sheets. I talked to a fair amount of people after the exam, and they all said that they were able to do well only because the exam questions were almost carbon copies of the TA review session questions and that they put them onto the cheat sheet.

I'm sure this class was much better with 30-200 people. ECE C143A and 102 supremacy.

The professor and TA's are AMAZING, and I learned a lot from this class. However, words cannot explain how severely I disagree with the syllabus and logistics of this course. I think it was a bad idea to have a midterm be worth 50% of your grade. I understand that they changed it from 30% to 50% to even out the distribution or whatever, but it doesn't really make sense to up it by 20%. I feel like 40% would have been a lot more fair. Other than that glaring issue, this course is great and Prof Kao is extremely helpful and kind and approachable. I'd recommend this course to anyone interested in ML.

I imagine this class was much better in the past, but this iteration was just awful. Unenforced prerequisites led to a massive class (over 400!) of folks who weren't familiar with linear algebra/matrix calculus asking questions about course prerequisites NONSTOP, and Prof Kao having to explain these. We ended up falling behind by 8 lecture hours (4 lectures of 2 hours) compared to last year's iteration because of this.

Additionally, it seems Prof Kao experimented with asking more open ended exam questions this year compared to the last few. This is totally fine, except he added a closed solution rubric?? Exam average this time was low 80s (worth 50% of the grade => the grade distribution will be MUCH lower this time), and led to Piazza being spammed endlessly for students seeking additional 1.5% extra credit (no, I don't want nor need an Excel spreadsheet to calculate my course grade; stop posting this on a Q&A board). This is not how you run a class, this was a mess and easily one of the biggest wastes of time for anyone who has taken M146.

If the department doesn't enforce prerequisites nor cap the class size next year, DO NOT TAKE THIS.

Kao is great! Learned a lot. HW's can be tough/time consuming, but mosty because numpy doesn't always behave the way you expect. Dealt with influencers trying to disrupt class very professionally. Generally just a happy guy that seems super down to share what he knows.