Comments for LingPipe Blog

Comment on Convexity of (Root) Mean Square Error, or Why Committees Won the Netflix Prize by Viviane

Viviane — Sat, 04 Feb 2012 20:56:48 +0000

Thank you Felipe. I have read your work. It is ineedd the first to my knowledge that uses the extended (also called Fisher’s non-central) hypergeometric distribution for IR. I found that sampling from extended hypergeometric becomes exponentially expensive as the sample size increases, and that there is ongoing research in statistics about .While researching on the use of the hypergeometric distribution for IR, I found a paper from Wilbur dating back to 1993! Wilbur models the vocabulary intersection between the query and a set of relevant documents using the central hypergeometric distribution. Little has been done since then, probably because the multinomial distribution is a good approximation to the hypergeometric for most IR scenarios, i.e., when the sample size (query) is cosiderably smaller than the population size (document). However, as we show in the paper, in the case of document-long queries, the multinomial approximation does not hold anymore, and the use of the “vanilla” hypergeometric distribution is required.

Comment on How to Close a LinkedIn Account with a “Large Network of Connections” by Bob Carpenter

Bob Carpenter — Thu, 26 Jan 2012 20:12:22 +0000

@Aleks: 1. I’ve never had a problem keeping in touch with people I want to keep in touch with. 2. It was exactly all this notification stuff I was tired of seeing (though Aleks just told me in person you could turn it off, though somehow opting out altogether hasn’t even turned it off; yes I could fiddle with my own mail to do that, but that’s too much like work). 3. Ditto.

I also don’t have a Myspace or Myspace-clone account. I don’t even have a cell phone other than when I’m traveling. But I really like people in one-on-one or party/conference settings. And I’m pretty responsive to e-mail.

I just really really hate being interrupted. And I luckily have a personal and work life set up to make it possible (i.e., no kids, long-term project work, no production system maintenance, etc.)

I do understand this is a matter of preference.

Comment on How to Close a LinkedIn Account with a “Large Network of Connections” by Aleks Jakulin (@aleksj)

Aleks Jakulin (@aleksj) — Wed, 25 Jan 2012 15:22:42 +0000

I derive three things from LinkedIn: 1) be able to keep in touch with folks as they change jobs and email addresses 2) I get notifications when people change jobs & positions (saving the boring conversations about “what are you doing?”) 3) an occasional introduction

Comment on How to Close a LinkedIn Account with a “Large Network of Connections” by Bob Carpenter

Bob Carpenter — Mon, 23 Jan 2012 19:58:21 +0000

Thanks for sharing.

I know it can be useful for the kind of thing you mention. But I’m not looking for any of that.

I didn’t find it helped our business at all. It could’ve helped me find a non-academic job.

I don’t find getting academic/technical books published that hard.

Comment on Bob’s ML Meetup Talk — Stan: A Bayesian Directed Graphical Model Compiler by Bob Carpenter

Bob Carpenter — Mon, 23 Jan 2012 19:53:05 +0000

We know HBC pretty well. I spoke to Hal about it at length before starting the Stan project.

We’re pretty up on all the competition. More like what we’re doing are the recent Church extensions that do HMC. Here’s the original paper:

http://www.mit.edu/~ndg/papers/churchUAI08_rev2.pdf

but I think there were more recent things at NIPS this year.

I also know the PyMC folks were working on similar HMC + auto-dif approaches.

Also, theano in Python is trying to do some of the same things:

http://deeplearning.net/software/theano/

Comment on ThinkPad W510 Review: First Impressions by Andrew

Andrew — Sun, 22 Jan 2012 06:40:37 +0000

Andrew…

[...]ThinkPad W510 Review: First Impressions « LingPipe Blog[...]…

Comment on How to Close a LinkedIn Account with a “Large Network of Connections” by Steven Groves

Steven Groves — Thu, 19 Jan 2012 00:55:20 +0000

Bob,

I’m not a Linkedin guru or anything, but I will share an example of how Linkedin connections helped me in the past. To be fair, I am more of an online / offline guy than an offline / offline guy, but nonetheless.

I had never planned to be an author, but while I was working with Lon Safko and the SocialMediaBible.com, we received a call from a market researcher in Atlanta who wanted us to provide some input on a tactic they were considering. We took the call and talked a bit and afterwards I looked up the caller on Linkedin. Turns out he had already published a book previously on marketing ROI and social media in 2009 was still looking for an ROI. We talked more and agreed to collaborate on a book called ‘ROI of Social Media’ – while not on the best seller list, it is widely read by marketers all around the world.

Fast forward to late 2010 and another contact referred me to a listing on LinkedIn that led me to a project which covered all my moving expense to Colorado from Arizona, something again I wanted.

The bottom line is that Linkedin might be about finding a job for some, but for most of us on LinkedIn it is about the contacts and networking for business – not leisure / personal like Facebook. for me it led me to my first internationally published book with a named publisher (John Wiley & Sons) and to a move that I wished for.

Almost daily I find uses for my professional network in Linkedin that is mutually beneficial to myself and my network.

Well, that’s my story – yours certainly could be different, but I’ve found online social networks will pay you back what you put in.

Socially Yours,
Steven Groves
http://www.linkedin.com/in/StevenGroves

Comment on Bob’s ML Meetup Talk — Stan: A Bayesian Directed Graphical Model Compiler by Jose

Jose — Wed, 18 Jan 2012 05:34:34 +0000

You might also want to check Hal Daume’s HBC (Hierarchical Bayes Compiler).

Comment on Probability Measures and Random Variables by Online Statistics Course - Introduction to Random Variables

Online Statistics Course - Introduction to Random Variables — Mon, 16 Jan 2012 11:05:21 +0000

[...] #3D3D3D; } ul.dates .date{ color:#858585; padding:0 1.5em 0 0; } // Networking CommunicationsProbability Measures and Random Variables [...]

Comment on ThinkPad W510 Review: First Impressions by product reviews

product reviews — Sat, 14 Jan 2012 03:15:15 +0000

product reviews…

[...]ThinkPad W510 Review: First Impressions « LingPipe Blog[...]…