align-toparrow-leftarrow-rightbackbellblockcalendarcamerachatcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-crosscrosseditfacebookglobegoogleimagesinstagramlocation-pinmagnifying-glassmailmoremuplabelShape 3 + Rectangle 1outlookpersonplusImported LayersImported LayersImported Layersshieldstartwitteryahoo

Sparse and Low-Rank Recovery Problems in Machine Learning

Speaker: Jeremy Watt

Title: Sparse and low-rank recovery problems in machine learning: an introduction to applications and models

Abstract: Due to their wide applicability, sparse and low-rank recovery models have quickly become some of the most important tools for today’s researchers in machine learning, statistics, optimization, bioinformatics, as well as signal, image, and video processing. In this introductory talk I’ll focus on motivating sparse/low rank models from the natural desire to tweak a) standard linear least squares regression and b) Principal Component Analysis (both of which I’ll review briefly first) in order to perform a variety of machine learning tasks on big data sets.
I’ll also discuss (with pictures and live code!) an array of amazing applications of these models including: lasso problems in biostatistics, face recognition, digital compression, Robust Principle Component Analysis, Collaborative Filtering (e.g. the Netflix Prize), compressive sensing, as well as a number of fascinating recovery problems in image processing.

Bio: I’m a PhD student at Northwestern working in the rapidly growing field of sparse and low-rank models and algorithms, with applications to machine learning and signal processing. You can find out more about this exciting field of research - including explanatory notes and presentations on machine learning topics - on my blog here. I’ll also be co-teaching a course on the subject of this talk during the upcoming winter quarter – for those interested (all are welcome) a tentative syllabus may be found here.

Join or login to comment.

  • Ganesh

    1000 character limit - baaaah.

    Is it possible to map other problem domains to a frequency spectrum sparsity? It possibly could be but it is not the intuition I came out with after the presentation. Perhaps my brain is too dense (un-sparse - get it?) to be able to get that intuition.

    The discussion about contrasting the nyquist criterion as a sampling limit with the sparse summation of sampling terms was quite interesting.

    September 5, 2013

  • Ganesh

    It was a good presentation and whatever was presented was materially apropos to the scope that was clarified in the first 5-10 minutes of the presentation. Ready.Set.Go - Gottsta check it out!
    I do have some cognitive dissonance about heavy reliance on image processing/filtering etc. (may not be the right technical terms) to talk about sparsity. It seems like (at least from the presentation) the contrast between "the natural" aspects of an image (with its affordability to do fourier summation of sparse number of terms on the frequency dimension/spectrum) and the unnatural/artificial superposition of digital text layup (all red in the example - so easy to filter it out from a frequency spectrum point of view) - seems like it makes a "natural problem domain" for sparse matrix approach.

    September 5, 2013

  • kyle

    Are there any materials to look over? I had to work late on the day.

    September 4, 2013

  • Gene L.

    I thought this was an amazing meetup overall. It was very compelling, understandable, and the theoretical background was appropriate.

    I feel like some of the hard details were glossed over, and it made the sparse matrix approach seem a bit like a magic wand... I kept thinking about where the catch was and there were some catches.

    Also, I wish that Jeremy had been a little more self-promotional. I was wondering what his current work is and why he works on these problems. Also, when he mentioned that he's "doing a course" I wanted to know more about the name / cost of the course (although I see the link is in the meetup).

    September 4, 2013

  • Zhiyong W.

    It is a great talk, Jeremy. Will you share your slide somewhere, like slideshare?

    September 4, 2013

  • BirmSoo K.

    I appreciate Jeremy for his excellent presentation for the introduction to the Sparse and Low-Rank Recovery Problems.

    September 4, 2013

  • A former member
    A former member

    Jeremy was a fantastic presenter of the subject matter. Given the range of people with varying levels of expertise, his emphasis was great.

    IMHO, the purpose of this meetup is not to delve into the details of any particular subject but to merely to whet appetites.

    I appreciated Jeremy's reference to Northwestern's Ready-Set-Go program. Even before you mentioned it, Jeremy, I was already thinking your presentation was among the clearest of this kind of subject matter I'd ever attended.

    Kudos to Jeremy, Rob, and Thomas for making this event happen!

    September 4, 2013

  • Yongheng L.

    Jeremy is a good presenter, but I personally feel today's talk focused too much on breadth and not enough on depth.

    September 3, 2013

  • Michael S.

    My first meetup, and so I didn't get an RSVP in, but hopefully 'one slot available' as of 11 AM on Tuesday means I will be allowed to show up. I'm looking forward to it.

    September 3, 2013

    • Rob L.

      Hi Michael: I sent the list to our host around 10 a.m. this morning, shortly after the RSVP deadline. So, unfortunately, I won't be able to add you. Sorry.

      September 3, 2013

    • Michael S.

      Understandable: I should have looked into it earlier, my fault. Well, if I can't be there, I hope that some of the slides and whatnot make it onto Jeremy's blog/the group. Thanks for the clarifying reply.

      September 3, 2013

  • Rob L.

    Hi all, don't forget that our venue tonight is NOT our normal one. Sapient will be hosting us at their location on 30 W Monroe. See you there!

    1 · September 3, 2013

  • A former member
    A former member

    sorry out sick today

    September 3, 2013

  • Miles R.

    Looks quite intriguing!

    August 16, 2013

  • Christopher S.

    Excited to hear another presentation from Jeremy! Enjoyed the generalized k-means one last time. Sounds like a great topic as well.

    August 16, 2013

Our Sponsors

People in this
Meetup are also in:

Start the perfect Meetup for you

We'll help you find just the right people

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy