| Category | Difficulty |
|---|---|
| HW | 3 |
| Exams | 6 |
The course is taught by Professor Byron Yu. The course focuses on applying machine learning techniques to neuron data. This can be an excellent companion course to 18-661 as the two largely mirror each other in content for a good portion of the semester. The homeworks are a mix of probability and MATLAB processing of real data from research on monkeys. It can be very cool. The homework is average in length, but MATLAB portions depend on the student's experience level. They can be completed quickly if students have background in MATLAB and probability.
Prof. Yu spends the first 3-4 weeks of the course reviewing how nerves work. This is interesting and not graded. After this point, homeworks are due roughly every week and a half. The exams focus on mathematics and are normally slightly more complex than homeworks. The class heavily uses Poisson processes. While knowing anything about them is not a prereq, any probability or statistics background will serve you well.
Some of the main topics covered here are
- Boosting, Random Forest
- Clustering
- Expectation Maximization
- Principal Components Analysis/Factor Analysis
There are ~8 HWs, a midterm and a final. Two homeworks are allowed to be turned in up to 24 hrs late.
Prof. Yu's office hours are very helpful. I took the course remotely, but they were still good. I think they were the most normal-feeling office hours of any remote office-hours I've been to. He is good about having students explain questions to each other and helping students work through problems. Start homeworks early so you can make it to the office hours.
Reviewing homework is definitely the way to go. Look through the examples Prof. Yu does in class as well as the homework solutions. Normally, exam questions are similar to homework but with a twist. Try to predict the twist that he could add and you will be more prepared.
You can check out the course here: https://courses.ece.cmu.edu/18698