Video 1 questions and feedback


(Aseem Bansal) #1

Hi @rachel

I know that the general pattern that you follow is some things are given high level and then delved deeper later. So if any of the questions are covered in later lesson don’t bother spending time answering just say skip till end of lesson X. I understand that getting answers for some of the questions would be possible to do via google searches but I don’t want to do that in case there is a more practical explanation coming up in the course. In case not please just tell me to google it.

Questions

  1. Why is breaking matrices into smaller matrices an important thing? I did not understand the explanation that was given in the video.
  2. What is Markov chains?
  3. It was mentioned that matrix multiplication is important because we can do things by hand and we can also do them at large scale. Isn’t that true for every operation? What makes matrices so special?
  4. What is eigen values?
  5. What is scanlines?
  6. By the end of lesson 1 what should have I learnt/understood/got intuition on? Asking this because of the way the course is taught sometimes it is hard to know if you really got to what the instructor wanted me to understand. So I think explicitly mentioning them would be helpful

Feedback

  1. When video is playing and you start speaking it is best if you pause the video. I wasn’t able to understand what you were saying at those times. I know you cannot change these videos but jut for future reference that would help a lot for the online video format.
  2. When speaking don’t look towards the board because the mic is not there. It becomes hard to understand. I was able to go through it because youtube has subtitles. Or can get a portable mic which could make things easier for you while not worrying about online videos while taking your lectures maybe.
  3. Thanks for making these videos freely available. @rachel you and @jeremy are doing great stuff. Looking forward to complete this and then restart deep learning part 1. My expectation from this course with which I am starting this today is that I be able to better read and understand the suggested reading material for Part 1. Not comfortable enough to be able to explain the stuff. Hopefully after this I am able to get there.

(Rachel Thomas) #2
  1. Why is breaking matrices into smaller matrices an important thing? I did not understand the explanation that was given in the video.

There are a lot of practical applications to decomposing matrices: topic analysis, data compression, Google’s PageRank, and more, that will come up throughout the course. Seeing the applications in the later lessons will be helpful in understanding why breaking matrices into smaller matrices is useful and common!

  1. What is Markov chains?

Markov chains are a way to represent probabilistic systems with different states. They show up in a lot of different applications/industries. You don’t need to know about them for this course, although for people who have already seen Markov chains, it’s nice to know how they relate to matrix powers. You can google this if you want more info.

  1. It was mentioned that matrix multiplication is important because we can do things by hand and we can also do them at large scale. Isn’t that true for every operation? What makes matrices so special?

Matrices are special (and important) because so much data is represented by matrices: including images, language, and anything you can put in an Excel spreadsheet or SQL table. Matrices show up all over the place in data science, so it’s important to be able to work with them efficiently.

  1. What is eigen values?

They will be covered later in the course.

  1. What is scanlines?

These are talked about in the Halide video (although aren’t used elsewhere in the course). I included the Halide video primarily to illustrate:

  • there are different choices in how a given matrix operation is implemented
  • these choices impact speed and memory usage
  • these different approaches have trade-offs
  1. By the end of lesson 1 what should have I learnt/understood/got intuition on? Asking this because of the way the course is taught sometimes it is hard to know if you really got to what the instructor wanted me to understand. So I think explicitly mentioning them would be helpful.

This lesson was intended to give you an overview of some of the

  • types of issues that arise when dealing with matrix computations
  • there are different considerations for computers doing matrix computations than when doing them by hand
  • matrix applications are everywhere