- 1 review
- 1 completed
This is by no means an easy course if you want to do more than the concepts stream. But I was glad it wasn't made out to be an easy course with an easy way to certification. The course was run concurrently for UMN students as well so you've got to take into account that this is a university grade course. The video lectures were long as some people have commented but there is simply a lot of material to cover. I didn't find it slow because it is interesting to understand some of the background and the problems that pioneers encountered. I liked the fact that the course was not dumbed-down. I also liked the wide variety of interviews with other experts and practitioners in the field. On balance, I think the course worked well. There were some execution glitches for the peer grading work - we needed more guidance as to the nuances of a poor, mediocre, good and excellent answer but that part should improve if they choose to run the course again. I did the NLP course by Columbia University earlier in the year and in comparison, I think they are equal in terms of learning outcomes and difficulty but with different approaches. With the NLP course, your mark is based on the level of improvement in the NLP models you were building. The recommender system course has a more traditional approach to assessment.