Professor Sankararaman is an amazing lecturer, and he explains concepts in a clear and concise manner. For my quarter, there were eight quizzes in total (pretty much one per week), which was somewhat inconvenient. However, I learned a lot from this class, and it was overall a highly enjoyable experience. I would definitely recommend taking this class with this professor.

grades were curved down. homework 50%, midterm 20%, and final 30%. if ur exam scores on midterm and final are higher than the median, and did all the homework, u should secure a B+ at least. not a easy class if u have no prior knowledge in ML.

Didn't attend a single lecture but the class is popular and well-structured. The TAs are very knowledgeable and very quick in grading. ML concepts are boring though.

echo the rest of the reviews this man is the goat

This is an amazing class. I can't recommend Professor Sankararaman enough. He is able to distill complex ideas into easy-to-understand and interesting lectures. His slides are slick, clear, and thorough. He also posted lecture videos online our year. However, I highly recommend going to class because it is really easy to fall behind if you rely on just the videos. This class is quite difficult. If this is your first machine learning class, you will have to put in a significant amount of effort to truly understand the material and get an A. Before the class starts, I recommend going over your Math 32A and 33A notes. You should be comfortable with multivariable calculus and linear algebra. You should also have taken a proper probability and statistics course beforehand. The projects and homeworks are pretty interesting. You'll be exposed to many different ML models and techniques such as decision trees, linear/polynomial regression, SVMs, PCA, boosting, HMMs, and clustering.

Highly recommend this class for those wanting a better mathematical foundation in machine learning and knowledge of the basic algorithms. The homeworks are mostly math problems and proof related to machine learning concepts, and usually the last problem involves programming one of the algorithms you learned in class. The tests are pretty much the same, minus the programming parts.
Sriram was a fine professor. He could be a little eccentric and sometimes go too quickly with the concepts + math proofs, but for the most part he and his TAs did a good job of making sure you could understand the concepts and the homeworks. Overall I recommend him as a teacher.

Siriam is an awesome professor. The class is very well-organized. There are several TAs who each hold lots of office hours throughout the week. The only complaint I have is grades were curved down at the end of the quarter.

This is a great class to take. The concepts are covered very well and the tests and homework’s are very fair. The material gets harder after week 5 as you do kernels and SVMs, so make sure to keep attending lecture. The grading scheme is tough so make sure you don’t lose points on the homework

Take it. The professor is very passionate about teaching and give clear instructions on what is going on every lecture. The slides are not very creative but they are clear enough even for self-study. The homework and problem sets are fairly assigned and graded. This class does not involve a lot of coding, so it is one of the easiest CS upper I have ever taken. He will give skeleton code, and all you have to do is to fill in the "to do" parts according to instructions, where everything is done in python. For lectures and exams, I feel it is more math and stats focused, but with the foundation is Stats 100A or other equivalents, one will be fine on those stuff. Honestly, easy A for a CS upper.

Lectures are clear and slides are provided.

There are easy quizzes most weeks.
There are easy problem sets with coding sections.
The final was easy compared to the EC ENGR C147 midterm (took in same quarter).
C147 helps with M146 more than M146 helps with C147.
Machine learning topics aren't viewed from a big-picture perspective and aren't really related to each other from week to week.
Machine learning topics are pretty boring.