Thursday, January 06, 2011

CS: Teaching Compressed Sensing (Part 1)

I was recently asked to teach compressed sensing to a crowd of buddying engineers. This is an interesting proposition as it parallels some of the questions I am getting directly in my mailbox, namely,

"Igor, I read some of these papers and watched some of the videos in the Big Picture and still don't know why I am doing this or how to apply it to my field". 

The fascinating part of this exercise is really to think of how to deconstruct most of what I know into something that can be easily parsed to the person I was four years ago. Presentation and teaching evidently include some showmanship but what becomes important is synthesizing some of the concepts down to the point where most of your audience could listen in for ten minutes and have their ahah! moment. Here are the first draft  slides of what that course would look like:.

Compressed Sensing: What is it good for ? (a course on inimaginable wonders and how to make them work for you)

I am wondering if this could fit in a ten minutes video a la Khan Academy.

7 comments:

Anonymous said...

The slides look good - they touch the most general concepts in a way that makes a lot of intuitive sense.

I think there's no need to have that much pictures in the though -- they always manage to detract me from the actual content, if not chosen carefully. But that's only my opinion.

I'm sure a lot of people would be really grateful, if you posted the video of your lecture on the net. I know a few undergraduates who would, for one.

Anyway, thanks for sharing your work.

Igor said...

Thanks for the feedback.

Igor.

Leon Palafox said...

I remember I had my Aha moment when reading about the demo of the single pixel camera, at least lets you grasp the essence of it, and most of my Aha moments with algorithms come when I understand an specific application of it.

Igor said...

Leon,

What explanation ? the one given here : http://nuit-blanche.blogspot.com/2007/07/how-does-rice-one-pixel-camera-work.html
or another one given in some presentation (in the affirmative, which one?).

Cheers,

Igor.

Anonymous said...

I do not understand the slides with "Within an infinite number of solutions some of these solutions are:". I understand the fractal slide, but the others I'm clueless. I think these need some explanation. Or perhaps I have no sense of humor or I am not very intelligent.

-JL

Igor said...

The example is really there to say that out of an infinite number of solution, you also have an infinite number of ways to choose the solution you prefer. The adjectives I am using are there to show that eventually you choose only the ones that have a feature you care for: sparsity being one of them, some people might provide some definition for what a cute solution is and so forth... Maybe I should make that clear in the presentation. Thanks for the feedback.

Anonymous said...

can this thing be implemented on Matlab or some other software,or it is still in pre stages?

Printfriendly