Thursday, April 30, 2009

Free New York: Bath and Gym


NY city had free public bath houses. See the picture on the left, today it is a studio. NY times has this article showing the genesis of these public baths going back to 1897.

On being (nearly) free, I saw a man riding the subway late night (after 2 AM). He would be seated as the train pulled into the station and people got off or on. Then as the train pulled away from the station, he would get up and do pullup and chinup exercises using the handle bars in the subway car (straps having disappeared long time ago). So, that is a $2 gym trip.

FOCS For Real?

There has been an ongoing attack on FOCS for some time. More precisely, on FOCS sites. Here are the real sites (on the left) and the "attacked" sites (on the right): can you spot the diff?. Thanks to Phil Klein for battling the squatters.
2007: http://focs2007.cs.brown.eduhttp://focs2007.org/
2008: http://www.cs.cmu.edu/~FOCS2008/http://focs2008.org/
2009:http://www.cc.gatech.edu/focs2009/http://focs2009.org/

Labels:

Wednesday, April 29, 2009

Bidding on Configurations

I had this thought that internet ad auctions could be more general and let advertisers directly bid to determine the configuration in which ads may be shown. I did a writeup, proposing an auction for configuration bidding in sponsored search. If I had huffed, puffed, I could have done more.

[In my youth, if I did anything at all, I would set a goal and however silly or difficult the goal, I would go for it, burst my lungs may be, spit out my guts perhaps, not sleep even, do whatever, but get there and then collapse. I lived life like that. Now I am still in my youth (not yielding an inch there), but don't seem to live that way, just less huff, less puff. Qué lástima!]

Labels:

Annoucements: Jon K, Twitter, Google Research

Jon Kleinberg gets the ACM-Infosys award. Google Research tweets.

Labels:

Sunday, April 26, 2009

Conference Culture Contd

Increasingly, conferences award best papers/posters. (I heard that) WWW09 awarded the best paper and poster awards to resp:
Congrats to all! Further, perpetual conversation about conferences (lance, noam) continues over dinner: WWW09 has Prof Dame Wendy Hall giving a dinner talk about "A Tale of Two or Three Conferences". She asks, "Why do we make life so difficult for ourselves by constantly spinning off new conferences? Why indeed in this digital age do we still feel the need to physically attend conferences?"

Labels:

Tuesday, April 21, 2009

Internet Auctions and Traditional Media

What does it take to build an auction system in real life to sell ads? You have to know auction theory and be able to apply it of course. In addition, you make hundreds of decisions, some large, many small, that eventually become the system. A great example is a draft writeup on Noam's publications page that describes an auction-based system for selling ad slots on TV (traditional media!). At the heart of it is a simultaneous ascending price auction for combinatorial goods, but the angst of what to optimize and tradeoff is the crux and explained nicely in this paper.

ps: WINE09 website is up. Submission Deadline: July 20th, 2009. Notification: September 10th, 2009. Conference: December 14-18, 2009, at the Eternal City.

Labels:

Monday, April 20, 2009

Streaming Slides

There was a DIMACS workshop on streaming, that tried to bring together not only sparse approximation, signal processing and theory CS researchers, but also Statisticians, Coding and Communication Theorists, Compressive Sensors/Samplers, and others. The organizers have managed to get the speakers to upload slides, enjoy!

Disclaimer: Andrew McGregor and others did a great job of pulling the workshop together, I was merely a fly on the wall.

Labels:

Saturday, April 18, 2009

Research and Manage

Researchers sometimes become managers. I am not talking about being a professor, for to be a professor is to be a startup company on your own, managing funds, space, students, courses, colleagues, committees and more all by oneself. I am talking about being a manager in corporate research labs, managing research equals.

An example I watched closely was Joan Feigenbaum when she managed a group at AT&T. One of her great qualities was to think big, create a foundation and develop a research area into an ecosystem for years. She seized the initiative on massive data sets and with initial results, got many of us into the data streams area. About the same time, she and her coauthors developed secure approximations that several have since explored. Also, I remember her angst formalizing clean game theoretic questions in IP networking; with talks and writeups, she got many to follow.

Others manage and excel, in similar or other unique styles, and Joan herself has managed and led in other areas and roles. We need ways to recognize such contributions, beyond mere blog posts.

Labels:

Ride and Work

A weekend taken hours early,
sunny friday afternoon stolen.
I went for a ride on a mountain bike in Manhattan. Starting from the center of the Island, ie., the village, that would be a ride with the first stop at the Little Red Lighthouse on Hudson. Then, few "fallen" fences will let you ride along the train tracks, take a dirt road to Dyckman Street and on to a northern tip where there is a tree leaning over the river.

Of course, if you are a researcher, you can't steal a moment even if you want to. I spent most of the ride planning the talk I will give at COCOON. I have been giving talks that present overviews of sponsored search problems, with a little dive into an auction or an optimization result. This time I decided to give a technical talk with a deep dive into one result, now I need a few more rides to figure out the result.

Tuesday, April 14, 2009

Sideways

Researchers sometimes argue sideways: "Sponsored search is an important problem. Last year it generated ZZ billions of revenue. I am studying the offline complexity of some vague problem in this area and will show NP Hardness...." I know that is a silly example, but we should ban any paper that quotes the revenue of a business in the Intro. More seriously, I dislike it when theory papers argue as follows: Sentence A, Sentence B. Sentence A is something impressive about a business in $'s, or users or running time or whatever. And Sentence B is the problem they study vaguely in that orbit and their result.

What is NOT Compressed Sensing?

Will someone (perhaps the phenomenally committed Igor Carron with his compressed sensing blog or the Rice folks) explain to the world what is NOT compressed sensing. It is NOT
  • Any sparse representation with small number of nonzeros.
  • Any dimensionality reduction by random projections.
  • Any operation that replaces L_0 optimization by L_1.
  • Any heavy hitters algorithm.
I do not want to listen to so many machine learning, signal processing, data mining, statistics, image processing researchers tell me that they do compressed sensing. Or may be I should throw up my hands and let the world embrace Compressed Sensing, after all they are sensing something, and it is compressed.

Saturday, April 11, 2009

Ray at Heart

Everything has many hearts, wears many hats. The pulsing Juilliard is one of NY's hearts, and near it, the Walter Reade Theater (WRT) needs only one hat, that of a movie theater. Beginning Apr 15, WRT does two weeks of Ray's films: Slideshow, Review: "simply his bottomless curiosity about how people negotiate the most urgent demands of nature and culture, is impossible to mistake, no matter what kind of Satyajit Ray movie you’re watching." You dont have to watch an occasional Ray's screening at MOMA, just grab a fistful at the WRT.

As WRT explains, "Born into a Bengali family in 1921, Ray was working in advertising when he met Jean Renoir, in India scouting locations for The River. Upon hearing a description of the story that would become Pather Panchali, Renoir encouraged the young cinephile to try his hand at filmmaking, but it was not until Ray saw Bicycle Thieves in London in 1950 that he resolved to become a director." So, lives trace many trajectories, and we get to relive some of them.

ps: A friend's email this morning made me postpone my plans for the day and do this post.

Wednesday, April 08, 2009

Reviews

I always wondered what academic research would be like if we brought the style and quality of literary reviews to our work, our papers, talks, punctuations, and all else. Here is an engaging review of Co., a pizzeria, by Lila Byock in the NYer:
In the world of pizza, debates over crust can resemble sectarian disputes. There are the ascetics, who demand pies as thin and brittle as Communion wafers, versus the libertines, who prefer something they can sink their teeth into. If anyone can persuade the quarrellers to break bread, it’s surely Jim Lahey, of Co. A rookie on the New York pizza scene, Lahey is an expert on crust....

Inside, long communal tables are ideally suited for diplomacy. ...

... - but it’s hard to quibble with the crust. Thin but not dry, crisp yet springy, it offers something for everyone—a middle way that doesn’t feel like triangulation. When the definitive history is written, Co. may be remembered as the site of the Chelsea Pizza Accords.
To those who need more on pizza, there is more. Me, I have been asked to keep away from these beasts. Doctor's orders.