paper for 8/6 meeting?

[Matt Lease]

Yinon suggested last time he might have a suggestion for our next
meeting. Below is web page for what we've done far with some ideas
for other papers to read. Feel free to add to the list!

http://piratepad.net/xEp3fuygiq

[Yinon Bentor]

The paper I wouldn't mind reading and discussing is already on the list:

Learning From Crowds.
http://www.umiacs.umd.edu/~vikas/publications/raykar_JMLR_2010_crowds.pdf

The reason I think it might be useful is that it follows the same
themes as papers we've discussed recently, but addresses regression
problems as well as classification. Its applications are a bit
outside of what we've dealt with before: most of the case studies
relate to biomedical data and the annotators are still trained, except
for one classification task that builds on the RTE experiment in Snow,
et al.'s paper. It is a journal article and a bit longer and more
ML-oriented than previous papers we've read. If we choose this paper,
I'm willing to lead the discussion on it.

I'm sure that some people are tired of this theme and are interested
in other types of work, and if that's the case, we might want to look
at other types of things to read.

Another paper that has caught my interest lately and might be
applicable to others in the group is Google's 2009 paper describing
Pregel, a system for large-scale distributed graph processing that
mirrors MapReduce in some ways. The citation is:

Malewicz, G., Austern, M. H., Bik, A. J., Dehnert, J. C., Horn, I.,
Leiser, N., and Czajkowski, G. 2009. Pregel: a system for large-scale
graph processing. In Proceedings of the 28th ACM Symposium on
Principles of Distributed Computing (Calgary, AB, Canada, August 10 -
12, 2009). PODC '09. ACM, New York, NY, 6-6. DOI=
http://doi.acm.org/10.1145/1582716.1582723

This paper may or may not be outside of the scope of interest for
other in the group, especially because it might be too
systems-oriented for this crowd.

Any other ideas?

-Yinon

[Matt Lease]

Let's go ahead with Yinon's suggestion and read the Raykar paper next
Friday.

[Matt Lease]

Next Friday 8/20 we will continue our readng and discussion of the paper
below

Yinon Bentor wrote:
> The paper I wouldn't mind reading and discussing is already on the list:
>
> Learning From Crowds.
> http://www.umiacs.umd.edu/~vikas/publications/raykar_JMLR_2010_crowds.pdf
>
> The reason I think it might be useful is that it follows the same
> themes as papers we've discussed recently, but addresses regression
> problems as well as classification. Its applications are a bit
> outside of what we've dealt with before: most of the case studies
> relate to biomedical data and the annotators are still trained, except
> for one classification task that builds on the RTE experiment in Snow,
> et al.'s paper. It is a journal article and a bit longer and more
> ML-oriented than previous papers we've read. If we choose this paper,
> I'm willing to lead the discussion on it.
>

[Fang Xu]

Hi,

This is very good instruction about training in logistic regression, which will help to understand the JMLR paper.
http://www.stanford.edu/class/cs229/notes/cs229-notes1.pdf

  Best Regards

--
XU Fang  徐昉