This review ended up rather long, so TL;DR: this class is hard and confusing- do not take it!
Unfortunately, I strongly do not recommend taking this class. It was probably my least favorite class I ever had to take at Columbia (disclaimer: I just took it as a second semester senior so I may be biased, but I still think this is a very honest assessment). I wanted to avoid taking Machine Learning because I was trying to have a chill semester, but I wish I had instead.
Professor Pe’er is an extremely nice guy, which makes me hesitant to write a poor review, but I had a very bad experience with this class. The material is hard, but what truly made this class truly difficult is the lack of resources to learn from. Usually, I can rely on the lectures, class notes, or the textbook to learn the material for a class. However, I could not understand the lectures due to Professor Pe’er’s delivery and stuttering, the class notes were confusing and extremely difficult to study from, and there was no textbook.
Professor Pe’er has an accent, but it is the stammering and stuttering that made it almost impossible for me to understand what was going on in class. He also spoke through a microphone which further distorted his sentences. I’m not trying to exaggerate, but it took intense concentration and patience during lecture to make sense of each sentence for me (moreso than any lecture I’ve ever had). Eventually, as a second semester senior, it was just too much effort for me and I was mostly zoned out during class.
There was no textbook to learn the material from. Thus, I was left with learning from the class powerpoints. The slides did not contain full explanations of concepts and diagrams, so they were very hard to decipher. I mostly ended up googling topics and watching Youtube videos to try and learn the material. This class is a mix of computer science, biology, and statistics and the material is unsurprisingly very hard to understand.
The problems sets were long and extremely difficult (8 in total). Most of the questions were worded poorly and required extensive clarification on Piazza before you could understand them. Given that the material itself was difficult and confusing, they took me hours and hours to complete. After a couple problem sets, I just stopped caring and scribbled down whatever work that looked okay because it wasn’t worth the time stressing over it.
For the midterm, you had to answer 3/8 questions. Not surprisingly, it was hard as well- the average was a 64. It was open book/note, but I didn’t even use anything I brought in. Afterwards, he gave us an extra credit assignment where we had to solve 3 more of the midterm problems (full credit received half the points between your grade and the highest class grade). If you do take this class, I would just focus on learning on 2-3 of the main topics that will appear on the exam (BWT, HMMs, etc.).
The project was okay. I was worried our project was too simple because everyone else’s sounded complicated, but most groups didn’t come close to getting good results so don’t overexert yourself.
In conclusion, this was one of the most frustrating and challenging classes I took as an undergrad, and unfortunately it was during my last semester here. Avoid taking it, unless you are truly interested in the subject- just take ML!!!! Sorry Professor!