Wednesday, January 28, 2009
Sunday, January 25, 2009
Waiting for SuperBowl XLIII
|Baseball has moved out, basketball is in The House, but some are waiting for the moment when the two masses meet at the line of scrimmage and the game begins with a snap, others are waiting for the sacks, blitzes, touchdown dances, and yet others for the commercials. On the left is the perennially creative Pepsi.|
Saturday, January 24, 2009
Going Broad in Ad Auctions
My coauthors (Eyal Even-Dar, Yishay Mansour, Vahab Mirrokni and Uri Nadav) and I study how advertisers bid in sponsored search auctions. Advertisers have a set of keywords in their mind and have to figure out how much to bid on each, given constraints such as their budget, goals such as maximizing their profit or the clicks, and statistics such as what each bid will yield in expectation.
Close to reality, applied algorithms research kicks in. Auctioneers --- Google, Yahoo! and Microsoft --- let advertisers bid "broad" on keywords so their ads will qualify automatically for auctions even when users type variations of keywords. Advertisers can not cover every keyword variation in their bidding set explicitly, and find the broad match feature essential in practice. The difficulty is that a single bid now applies implicitly to several auctions, each with different value to the advertiser. Biding decisions now become nontrivial. We formulate associated problems and provide exact/approximate solutions in a forthcoming WWW09 paper.
Those who are interested in this topic may also want to look at this paper that studies the dual view of an auctioneer, ie., revenue, efficiency and equilibrium properties of broad match auctions (paper appeared in Workshop on Ad Auctions 08).
Monday, January 19, 2009
Data Streams in Barbados
Sunday, January 18, 2009
Wish List: How to write a reference letter
Thursday, January 15, 2009
Dance your PhD
Tuesday, January 13, 2009
It is just Tuesday and it already feels like it has been a long week, so here is some distraction for anyone feeling blue. For those who care, he says, "What's up? you arent gonna take a stance..."
Wish List: How to write a statement
Many of us review application packages from students abroad for graduate schools. Besides the scores and transcripts, students submit a Statement (of Purpose, of Research). These are typically very poorly written with flowery sentences, emotional claims about ones' obsession with computers, mathematics and the academic world, and definitive view of what was destined. Notwithstanding the material on the web with general guidelines and paid help, will someone write a short note on how (not) to write a statement for graduate school in CS? I like short sentences in active voice, north american style.
Sunday, January 11, 2009
The NYer continues to thrill me. The Jan 12th edition has many gems. One a quote from Hilary Clinton about Bill, "He's never met a sentence he couldn't fool with." Another, a review of the film Defiance about the survival of a group of Jewish people in a forest during World War II, says: "The farmers or workingclass men who could shoot, gut an animal, and build a shelter were sought out as protectors by the women, including the educated, upper-class women; the formerly desirable scholars of Hegel, Marx and the Talmud were not." It is a time of reversals.
Secretary of X
Outdated spams and outed scams
This scam originated may be one year ago, and has been outed. Then, why do these emails continue to circulate? Similarly, one gets spam comments on ancient blog entries that presumably hardly anyone reads. Is there non-zero utility in scamming via outed scams and outdated blogs?
Friday, January 09, 2009
Winding the week down: Mathematicians Score
|The Best||The Worst|
|1. Mathematician||200. Lumberjack|
|2. Actuary||199. Dairy Farmer|
|3. Statistician||198. Taxi Driver|
|4. Biologist||197. Seaman|
|5. Software Engineer||196. EMT|
|6. Computer Systems Analyst||195. Roofer|
|7. Historian||194. Garbage Collector|
|8. Sociologist||193. Welder|
|9. Industrial Designer||192. Roustabout|
|10. Accountant||191. Ironworker|
|11. Economist||190. Construction Worker|
|12. Philosopher||189. Mail Carrier|
|13. Physicist||188. Sheet Metal Worker|
Wednesday, January 07, 2009
ps: Claire Mathieu did a few tweaks to the SODA conference (the printed program had abstracts and proceedings finally in CD), and they really worked.
Saturday, January 03, 2009
The girl looked composed and determined and was clearly the one who had the idea to approach me. The mom was a little reluctant at first, and then increasingly engaged, as the girl looked on. It was the day of the first snow during my trip to Japan. It was the beginning of a three hour train journey through snow, mountains and tunnels into Hokkaido. It was the evening, sun more or less down, buildings grey, and the lights unable to do much more than be dull.
Slowly as the train started climbing, snow piling on tree tops, it occurred to me that someday in the future, she would be a grown up, she would have learned to speak English, may be she would have a bad day and remember a good moment, remember the sembe she gave a stranger who was studying the glass mural in the station, she will google "Hachinohe cookie" and would find this blog entry, dated but indelible on the internet.
People fret that blogs are stored forever, what if something innocent of today proves embarassing in the future. I fret too, but this post is more a wish that someone, a particular one, will someday discover what I say today.
More Compressed Sensing
It is always interesting to see the simple CS ideas reach out to new areas from centralized L_0 to L_1, to distributed CS, to devices and hardware, and now to secrecy. Yaron Rachlin and Dror Baron explore whether CS measurements provide secrecy about the underlying signals, and what can be salvaged.