- 7 reviews
- 7 completed
Most of the material in the first half of the course was at least somewhat familiar to me. My interest was in the second half covering transport, routing protocols, and network planning. This course is very comprehensive and immediately valuable to anyone who has ever tried working the material in a self-directed manner. This was, however, my first course at 2x playback speed. Read into that what you will. I enjoyed the survey of so many related technologies. Prof. Wetherall focused on the technical progression from the 70's through to modernity and properly motivated some of the protocol tweaking among big players (e.g. Google, Amazon). I now feel more prepared to tackle modern material which was my goal for taking the course. I found the carefully worded questions a bit annoying. This could be chalked up to poor reading comprehension on my part but I often ended up misreading questions. No worries though, easily correctable after reviewing auto-grader feedback.
Fun experience with a focus on applications. The Bayesian filter material is complicated but quizzes are straightforward. A less prepared student could come back to the material in his or her own time while still certifying in the course.
I expected an introduction to statistical techniques used in a fMRI setting and the course delivered on that expectation. The exposure to magnetic resonance physics and pre-processing was very enjoyable. It was very much an overview however and the inquisitive student would need to be very motivated in pursuit of a deeper understanding. The TAs were very supportive of that pursuit. I was hoping for some hands-on experience with actual fMRI data. There are probably logistical and maybe even ethical reasons why that isn't possible but it would have been very useful to me.
My prior experience is primarily with Voice-over-IP product development. My data networking skills, however, are pretty weak (scored about 60% in the preliminary/prerequisite exam). I was motivated to take the UW Computer Networking course by the disclaimer in the first offering of SDN. The Computer Networking course was, in some ways, more difficult for me. I found Prof. Feamster's presentation style very easy to follow. The Computer Networking course does not have a programming component. I didn't struggle with the programming component of the SDN class mostly because the preparation of the assignments was very thorough. If you carefully follow the instructions then you'll do just fine. I most enjoyed the interviews with industry and academic leaders. It is worth registering for the course for those interviews alone. Prof. Feamster asks good, if not occasionally leading, questions. It is a real pleasure to hear about the evolution of the technology from those most directly involved in creating it. I was most distracted by the emphasis on BGP. Prof. Feamster and his team are obviously most interested in applying SDN techniques to gateway routing protocols across the Internet. You will probably fail to appreciate the application unless you've struggled with the existing technology. You will, in other words, need to have professional network administration experience to appreciate the argument he makes throughout the course.
Time management is key to success in this course. Prior programming experience is mandatory, prior exposure to the field definitely useful. This course is relatively unstructured compared with others on Coursera. Some people found the lack of structure liberating. I'm probably a bit too used to being spoon fed material in the MOOC format. I don't think the format detracts from this course, it just didn't really fit my work flow very well. I really enjoyed the passion for the subject among staff. The focus on practical skills and "war stories" gave me a sense of glimpsing into the field. Some students rose to the programming challenge and that was also inspiring to read and work through. Pressed for a dislike I would have to say that the starter code was a little bit too trivial. Greedy was almost always the way to start each problem. The starter code could have been closer to that solution without giving away a lot.
Prof. Thrun has a really engaging presentation style. It helps that he is one of the foremost accomplished practitioners in the field with an exhibit in the Smithsonian no less. You know you're learning from the best. The course was almost exactly what I expected. I might have wanted a bit more "starter code" for my own project but that expectation seems naive in retrospect. Anyway, the problem solving techniques are the important bits. The rest is too project- specific for a remote learning environment. It is difficult to imagine how this course could improve. Some people found the "utilitarian" approach to the mathematics a little distracting. You are asked to accept a few results on faith but that was never too difficult for me. I might have added a few links for the mathematically-inclined. I haven't taken it but there is another Udacity course on mathematical modelling. It might be the one on differential equations. I did quite a bit of that in my classical education so it will probably be a while before I go back to it. Perhaps it is an alternative for you. Some proficiency in Python would be useful for this course. I would also consider the Khan introduction to Linear Algebra, just so you are not seeing a matrix for the first time in this course. That said, the emphasis is on mastery at Udacity. You might want to use this course to drive your linear algebra learning.
Prof. Klein is an engaging lecturer. I really enjoyed his enthusiasm for the material. I expected an environment similar to the other more recent MOOCs but edX is quite different and in a good way. The discussions were organized lecture by lecture, problem by problem and that seemed to focus the discussions. It certainly helped me to find the discussion(s) relevant to my questions. I've taken several AI classes: ML from Prof Ng, Robotic Car with Prof. Thrun, and Intro. to AI with Prof. Thrun and Prof. Norvig. Some of the material was familiar but the focus on adversarial scenarios in this course was fun. The others kind of survey the field. This one did survey but with an eye on how to program a game. The game was the running example and it was fun to think of the problems that way. You should probably have seen Python before taking this course. You really don't need any background in probability to grasp reasoning under uncertainty. High-school algebra and a bit of programming experience will take you a long way in this course.