First half of the subject was great. The lecturer (Chris I think?) was very good at explaining the content clearly. However the second half of the subject took a nosedive in quality with the new lecturer just because the probability portion just seemed a little too complicated. I know the subject used to have a Probability prerequisite before, but I still feel they hadn’t adapted the content well enough to suit CS students who hadn’t touched probability in quite a while like me. I found it very hard to follow and the lectures weren’t at the same pace as the first half, being much faster in pace. While revising for the exam I had many breakdowns due to just not being able to understand how to answer the probability questions, as there are lots of hidden tricks and derivations which you are expected to figure out using your problem solving skills. There is nothing wrong with that, but it does make problems a lot trickier. Another thing with the second half is that there are some more algorithms around arc consistency and other concepts that are taught, but the way they are taught in the lectures is slightly different to the way they expect you to apply the algorithms in the problems you are given. This was annoying since I only found this out after scouring the ed discussion forum. As for projects, both projects are pretty tricky when it comes to actually implementing them and doing a good job. The first project they wanted you to develop a path finding algorithm for a board game that they invented from scratch. It was honestly pretty tricky as even with lots of work and thinking my group weren’t able to pass all the test cases. Luckily there is also a report component, which meant that we were able to salvage some marks. For the second project, we were asked to develop a bot that could play this board game against other bots, and again, this was a tricky assignment to code and implement, but thankfully there is another heavily weighted report component which again meant good marks even though the code didn’t pass all the test cases. Both projects were tricky but I funnily enough found the first one harder, because it required more creativity and problem solving skill than the bot one. The tutorials I found were kind of bad. Even though I think I had a good tutor, I felt I barely had any time to talk with them and ask for help, since the class was decently big and there were a lot of questions, so they had to keep moving. The tutorial questions were often very difficult, especially the probability ones. It wasn’t fun when I was doing them for revision before the exam. Regularly I’d only really know how to do maybe 1 or 2 SUB questions on the sheets. As for the exam: 70% weighting for a COMP subject is pretty harsh in my opinion, especially with projects that required maybe 15-30 hours on average to do. The exam being in person and no notes meant it was quite tricky, since especially with probability you had to remember all the formulas off by heart. I found the non probability questions on the exam to be a mix of easy and strange, with strange meaning I had an idea of how to do it but was not sure what approach they expected. As for the probability questions, they seemed easy from first glance and quite mechanical, but when I started to actually do them, I realised that there seemed to be a lot of work actually needed to do them and finish them. Of course, it’s definitely possible so hadn’t thought of an easier approach, but I found them very difficult and left quite a few blank. I did end up getting a low H1 for this subject, and I think this definitely meant that the subject was scaled quite heavily. Considering there were quite a few tutorial sheets where I could barely even do a single question, I wouldn’t worry too much if the same situation happens, as scaling does seem to save the day. Overall I found this subject very difficult, and I was panicking before the exam as I did not feel prepared at all, and after the exam felt shattered. The first portion of the subject was quite enjoyable, and the projects while tricky can be fun to work on. However due to how much skill was expected for the probability portion, and how rushed it was taught, I can’t recommend this subject if you want good grades unless you have done probability to some level recently.

Pretty bumpy ride. I actually really enjoyed the assignments. Sure they're brutal and grueling and just straight up take up all of your time, but it's just so interesting and stimulating (unlike compsys which is only artificially difficult, and yes I just had to put that in here). The content itself isn't actually very hard. But man were the assignments brutal. I guess I was kind of stockholm'd into liking it. But seriously, if you're up for a challenge, and really want to explore and be ambitious, this might be it. That said, the Ed support is downright dreadful. I have no clue why they cherry pick posts to answer (like 2 out of like 15 posts every now and then towards the exam) and left the majority of it unanswered the way it is. It's honestly a nightmare, and very representative of the current state of this uni computing faculty. It's a bad, bad subject if you consider Ed support integral. And the tutorials are also a bit pointless. Essentially it's a discussion session with friends so that you won't forget what you study the last week. For the most part, it was not particularly helpful. Anyway, definitely the root of my pain this semester but I would still recommend it. It is poorly managed, and you most likely wouldn't get any exam support unless you join the consultations (but let's be honest, you'd most likely encounter a new problem eventually but hey, no one's there to answer anyway). Recommended with a big disclaimer of infuriation and disappointment.

Fantastic 3rd year subject. Sarah and Chris were clear and clearly showed passion throughout teaching which rippled through to my own enthusiasm to go with the subject. The content is interesting, especially if you loved the 1st and 2nd year algorithms subjects. The best part of this subject are the assignments, creating a game playing AI agent over two parts. You truly feel like your creativity is explored and rewarded in these assignments. The head tutor Adam was excellent in class, providing clear explanations and expanding on content from the lectures for our own benefit. Exam was just a straight rip of 2016 though I guess that's because of COVID. Regardless, one of the best level 3 COMP subjects you could take.

Really, really interesting content. Massive thanks to Matt for the work he put in as head tutor to make the project as good as it was. Why the exam is 70% is beyond me, though. I understand that having heavily weighted projects comes with challenges, and perhaps weighting the exam more heavily would be a valid fallback if people have issues with their team (the project is completed in teams of 2), but by default the project should be 50% at least. It's where all the learning happened, for me anyway.