ECE 4501

Not a very good first iteration of the AI Hardware class— the course was lazily put together and poorly run. I'd say ~70% of course content was clearly taken from other conference presentations, guest speaker presentations, or other slides from better-taught courses at other colleges with the same material. Not to say that instructors shouldn't pull from other materials, but compared to the other material, original content in the course felt very disorganized, since the TA would correct the instructor in real-time about material presented (because it was clear it had never been actively reviewed), or the material presented was clearly never presented before with wildly different levels of complexity, sometimes causing issues with running over the time in class.

Assignments were usually just tutorials for whatever hardware was present, but we only had two copies of each board for a class of 30 people. It was also the class' first time working with the hardware, so we had multiple instances where boards broke and only one copy would be available. All the assignments were peer-graded, but instructors failed to enforce rules about placing identifying information on the submissions, so none of the grading was done completely anonymously. Assignment grades also never released until the last week of the semester, well after any point where students could have feedback.

The project rubric was entirely unclear with no indication of what parts of the project would be weighed a certain amount of points, despite the fact that it was also 40% of our course grade. The project schedule was not established at the beginning, so starting the project was put off until the week before Thanksgiving, given each team realistically only two and half weeks to work with the boards. This made presenting what little results we could muster in the limited time we had at the end of semester awful, because this was our first presentation for the class, for a project worth 40% of the grade which did not have a clear presentation. Without any feedback or prior attempts, this whole exercise was essentially just punishing students for not having skills that they literally never had to chance to work on and improve in the class. Project grades of course also only released at the last week of the semester, well after any point students could contest them, and the rubric for the full project was never released. I don't mind if the class is not giving everyone A's, but to stake so much of the grade on such unclear and uncontestable standards is not educational, it's just lazy at best and in practice frustratingly unfair for everyone who could've done better if only they knew what they actually being graded on.

AI is cool. Working with the actual hardware can be fun and educational. And the instructors aren't bad people. This class is just really lazily put together and really poorly organized, and to be honest it left a really bad taste in my mouth. If you're an ECE interested in machine learning, maybe consider taking FoDA or ML for Image Analysis or ECE ML when it comes out, and if you're a CpE, then you literally have an entire other department's classes you could be taking that are way better (see: ML, AI, F1/10 Racing, RL...) . I'd thought I've never see the day when an ECE course actively have worse attendance than a typical CS elective but the class this semester takes the cake. Unless the department radically revamps the syllabus and the slides and the assignments and the project, this course in its current form is not worth taking.

EDIT: The rubric for the project was released. I thought the grading was fair but the fact that it was released so late still doesn't address the majority of my concerns for the project or the course as a whole. If they release everything earlier with an actual course schedule/syllabus (and address issues with hardware and course material), then I think this class' next offering is okay. But if it continues to be this disorganized, I personally wouldn't recommend taking the class.