The quarterly review — Q1

This is my first quarter of graduate school at UC San Diego, and as a part of my Master’s degree requirements, I took three graduate courses - Algorithms, AI(Probabilistic Reasoning & Learning) and Database Systems.

I found the quarterly system to be the very fast-paced and extremely time-deficient. I could perhaps have managed my time a lot better had I taken these courses in a semester calendar. I found each course to be challenging on its own.

I fell in love with the AI course on probabilistic reasoning and learning. The course was on probabilistic graphical models, hidden markov models, learning and inference, and reinforcement learning. I got to code some pretty cool stuff through weekly homeworks, one of which included path prediction in a maze with dragons. Almost every weekend was spent on my AI homeworks which was due every Tuesday. Algorithms made me feel like thrash and tear out my hair. I attribute most of my sleep-deprivation to this course, with the extremely challenging homeworks that not only involved figuring out solutions, but also proving that they are correct. I learnt how to think about a problem, and also validate my thoughts. It was a great course, but took most of my time in the quarter. Not to mention, I sat all night writing my homework and dozed off without submitting it.

If I could go back in time and change anything, I perhaps would not have taken Database Systems - a course I underestimated deeply. In comparison to the other two courses, this one didn’t have any weekly assignments, and a midterm and a final constituted 85% of the grade. Considering I did reasonably well in my undergraduate database course, I expected this course to be manageable in comparison to the other two. Little did I know this one would turn into a nightmare. I should have known open book exams can be devilishly tricky. This one was was open book, open laptop and even open database. With only three questions in the final and one of them constituting two thirds of the final exam points, in retrospect, it feels like a risky course to take. You either make it or break it. There’s no middle. However, I liked the course content broadly speaking. It covered query optimizations, failure recovery, concurrency control and incremental view maintenance, the last of which was the most fascinating. From my degree perspective, all of these courses count towards my requirements. Since I plan to specialize in AI/ML, this quarter was more about breadth than depth. Therefore, my subsequent quarters will be more depth-focused.

In retrospect, I don’t think I could have planned any better, but I definitely could have managed my time better. All throughout I felt I was patching several leaks in a pipe at the same time. I should have focused better and not underestimated certain aspects of each course. I learnt massively though and all of the sleep deprivation was totally worth it in the end.