Life is a Rabbit Pellet

Ramblings of a Zimbrindian's travels, life, and research.



Saturday, January 29, 2005

Jon Stewart sez


In today's environment, you can't judge a book by its contents. Thus says Jon in the middle of this clip, referring to the Condi hearings. Speaking of good clips - guess what? - I like the attack on the Fox reporter too.

One of my favorite Jon speeches was actually unconnected to the Daily Show. What can you expect from a commencement address that starts out with "I had forgotten how crushingly dull these ceremonies are." On the other hand, it does have some good advice: "Love what you do. Get good at it. Competence is a rare commodity in this day and age. And let the chips fall where they may."

Zimbos at the Australian Open


There are three Zimbabweans at the Australian Open. As usual, they are much better than doubles than at singles. I find it really annoying that the press, and even the tennis authorities, pay so little attention to doubles tennis when it's so much more interesting to watch.

Two of them, Wayne Black and Kevin Ullyett, won the men's doubles.

Wayne and his sister Cara were knocked out of the mixed doubles at the quarterfinals by Max Mirnyi and Martina Navratilova, who were in turn knocked out by Ullyett and South African Liezel Huber, who were in turn beaten in the final (tiebreak in the third set - ouch) by a couple of Aussies. Cara Black had a really bad Open; she didn't qualify for the singles, and despite being seeded second in the Women's Doubles with Liezel, went out in the second round. Eh, these things happen, Cara.

3d Virtual Sailing with Ellen M


Ellen McArthur is an awesome British woman sailing around the world alone trying to beat a very tough world record set last year by Francis Joyon. One of the many things new about her bid is its public outreach component. And on top of the "This is New" list is a good 3d interactive animation package that allows you to follow her race.

Monday, January 24, 2005

Momel, Colbert, Cosmus, Snowballs

Good progress with the pitch stylization algorithm Momel today, thanks to the prompt response of one if its authors to my plea for help. Must say, his C code is a lot faster than the corresponding praat implementation. And using snack is a joy!


Terry Gross had an interview with Stephen Colbert today. I didnt know he was a practising Catholic! Or that he was at Second City. Or that people actually had to practice diction. Or that he had a professional life other than The Daily Show. There's also an interview with him by FilmForce that makes good reading. Oh, and he went to Northwestern.

Much discussion with MS, RL & DR about converting Cosmus to a service provider for making visuals for scientists. They're interested, DR even suggested the NSF grant we apply for.

Good snowball fight. We're a bunch of weenies though, got tired after half an hour of missing each other. Other than the one Misha clobbered my nose with before I knew we'd started.

JL started a machine learning blog. So far so good...

What I would do with $200



The keyboardless keyboard is finally here.

Friday, January 21, 2005

Back!

No posts for a while... drat...

Had a LDA reading group on Monday, learnt a lot. Wrote documentation, should probably convert the matlab files to a full interface.

Tried getting JL's cover tree to work on windows, keeps crashing and not consistently enough for me to be able to test it properly.

Made a new webpage for the TTI Machine Learning Summer School.


People are using movies I made in the past! Max Tegmark used it in a astro seminar at MIT. Someone from there contacted me about later on, but a gmail bug prevented the mail from getting to me for a few days. (Btw, the picture here isn't one I made, it's the picture associated with his seminar on the page that you get if you click on it.)



Heard about someone getting a nail embedded in his head and then not knowing it was there. For six days. He has now had it removed. Cost: $100k. He doesn't have insurance, so donations from the public are welcome.

Friday, January 14, 2005

Basketball game ends at 5-2

Just read in the Chicago Tribune that a basketball game between two high school teams in Vermont ended 5-2. Presumably shot clocks will soon be introduced...

"We've been talking about it all morning over here," said Bob Johnson, the director of student activities for the Vermont Principals' Association, which governs high school sports in the state. "It had to have been one of the most boring games in the world."

Elektra

Going to see a bad movie today, after some convincing from Hunch that he knows it's a bad movie. It's got some great bad reviews, and a rottentomatoes rating of 2%. (In other words for every critic that liked it, 49 panned it.) I didnt know RT ratings went that low.

"Produces levels of excitement typically attained by proofreading science textbooks" - Sean O'Connell,
FilmCritic.com
: "

"There are a few laughs in "Elektra," principally because the script is a joke" -- Manohla Dargis, NYTimes

owowowowowow

The weather out here is not freezing. Freezing is 32 degrees Farenheit. Freezing is warm. It's ten degrees out there (I just got in at 3am) and that's before windchill brings it down another fifteen degrees.

I expect Mary Schmich is going 'whoopee!' now.

I aint leaving home tomorrow. The highest temperature for tomorrow is the temperature now.

Had a good meeting with John L et al about the TTI Summer School, and a short reading group about prosody with Gina and Siwei. Learnt stuff. Should have read more, of course.

Started learning Praat.

Made several nice mpegs for Stefano and Misha.

Thursday, January 13, 2005

Mamoca demo


Made a nice picture (right) for Stefano, a mechanical engineer who was testing some Laplacian Eigenmaps code of Misha that I rather heavily modified. He works on markerless motion capture (mamoca) (i.e. how to record body movements without having to stick sensors all over your subject's body) at the Stanford Biomotion Lab. Trouble is, I've helped him out a lot, and he'd like to cite me, but I dont have any relevant publications to cite... that's the problem when you fix up algorithms instead of inventing them...

On Language Barriers

Ananova, everyone's favorite source of quirky news stories (new stories! news Tories!) has a nice post on how a bank clerk foiled an amateur robber by pretending not to understand him. Even funnier, the title of the story is "Language Barrer (sic) foils robber". Good thing poor spelling isn't a language barrier.

Wednesday, January 12, 2005

Levels of Dislike


How you hate a person can be measured by what you would do if you were with them on a cliff:

  • You wouldn't tell them they were getting too close to the edge.
  • You wouldn't tell them they were getting too close to the edge, even in dense fog.
  • You would direct them to the edge in dense fog.
  • If they fell off, and climbed back up, you wouldn't give them a hand.
  • If they fell off, and climbed back up, you'd step on their hand.
  • You'd push them off.
  • You'd push them off, then push off a boulder after them
  • You'd push them off, then start an avalanche after them
  • You'd blow up the whole cliff, even if it took you with it.


permalink

Astro Visualization update

Darn, forgot to see Alan today. Better get in earlier tomorrow (say, before 3pm) and show him his group's data on Scope's geowall.

Mark and Randy seem to be having a whale of a time at the AAS Gadgets and Gizmos section - great! We're the hit of the session, apparently. Rather unfair, really, since some of the other demos are almost certainly more useful. We're just prettier. Mark tested one of Jason's students' games on a kid who wandered in, it seemed to like it. Anything involving shooting is good.


Got back in touch with Jess to see if she had any more data for us. This is after a break in comms of over ten weeks. Turns out she does, so there may be new data soon.

Daisuke, Brenda and I had a nice session getting Daisuke's data to look better. Good session, learnt stuff. Now using polygons instead of points... now I just need to figure out a way to subsample his data to make it smaller so we can get more timesteps.

Gave Brian a suggestion for how to deal with multiplanetary exosystems on the Digital Universe.

Added a shameless plug for Randy to his Astro Visualization Workshop on the main Cosmus page. He said I should submit something for it, we'll talk about that when he returns.

Dredged up the Galaxy Type Classification project thanks to Mark sending an email about a grey man wanting to see it. I tried PCA, Laplacian Eigenmaps, LLE, and Isomap on the same data (looks good when the field of view is 90 degrees - thanks Cart for that technique!), and sent it to him just now. He'll try it tomorrow. Vik is back in the equation.

Cart says I should apply to the NEI Focus Group at Chabot. Flattering, but that's not for me yet. I suggested to Mark that some of the UIC students who made geowall games for Adler should go instead.

On blogging science

Added a few entries to the SciTech portion of Wikipedia Current Events.

Such posts are harder than I thought. And is it worth doing so on WCE anyway? Or better done on Science Blog?

For example, what sources should one cite? Press Releases are good, since that's where the press gets its information from usually. How about the original journal abstract? Good, but not always understandable. NYT is good for a while, but after a certain amount of time articles become pay-per-view. BBC is good, since those links are pretty permanent. But how many good science articles don't get BBC pages? I'll have to look for those.

I debated whether to note that the first author on one of the articles - about the discovery of the three largest stars to date - is actually an undergraduate (nice work, Emily!) but decided against it. Figured I'd focus on the science rather than the scientists.

And should such posts include the scientists' names or nationalities? Dunno. I'm putting in nationalities to force myself to get a larger variety.

And should one blog science stories that are not well covered but aren't recent? Like the SALT telescope in South Africa, the largest single telescope in the Southern Hemisphere? (Note the word 'single'. Otherwise you'd probably want the Pierre Auger Observatory in Argentina.

I'm also pretty tired of science journalists only getting stories from Nature and Science and sometimes PNAS, or stuff announced at large meetings, so I'm going to look around for other stuff. Especially about speech and language related topics.

Sunday, January 09, 2005

Simulation Geowall demo done


Daisuke fixed the bug we suffered from data centering, so now we have a nice simulation of the evolution of part of the universe. The scale of the picture is about 50 million light years. Hopefully Randy and Mark will show this at the AAS Gadgets and Gizmos session tomorrow or Tuesday.

Saturday, January 08, 2005

Past Predictions of the Future

 
Jo just pointed me to a remarkably prescient article from a 1961 magazine that predicted Life in 2000. Since I have work to procrastinate, let me go through this point by point... and apologies in advance for the 1960s-style sexism present in the original essay. Thanks also to the Aussies who got the whole thing online!


What sort of life will you be living 39 years from now? Scientists have looked into the future and they can tell you.


So far so good, no definite predictions. No right, no wrong.

It looks as if everything will be so easy that people will probably die from sheer boredom.

True. For spoilt rich kids, the way to go is: Boredom -> drugs -> overdose.
For jocks: Boredom -> need adrenalin rush -> extreme sports (e.g. rock climbing sans carabiner).
For the rest of us: Boredom -> junk-food binging -> 'nuff said.

You will be whisked around in monorail vehicles at 200 miles an hour



Well done! This is true if you're Japanese or French or live in a country with high-speed rail service. Note how well the prediction avoids saying where you are whisked around. It could have suggested 200mph travel within cities, for instance.

...and you will think nothing of taking a fortnight's holiday in outer space.


Hmmm... pity they hadn't said 2020 instead (think Branson, not Shuttleworth). So let's play Fair & Balanced, and say "Absolutely true! People think of nothing involving taking a fortnight's holiday in outer space."

Your house will probably have air walls,

Wtf is an air wall? A wall with lots of air in it? Okay, score another correct. They don't say how much air, and lots of building materials, even bricks, store air as it helps with insulation. Several high-rise buildings have glass-air-glass walls. Budding prediction scam artists should note the fancy technique of using compound words with ambiguous interpretations.

... a floating roof,...

Obviously a referral to changing ceilings on mortgage payments. Or if you live in Florida, to the aftermath of hurricane damage.

,... adjustable to the angle of the sun,...

True. Everyone has roofs with daily adjustments to the sun's position. (Of course, it's the sun doing the moving, not the roofs, but as the earth is moving around the sun, so are the roofs, so this counts.)


Doors will be opened automatically...

True in shops. The prediction doesnt say it has to be true in people's homes. (Btw, the pic is from This is Broken by Mark Hurst. It was taken by Victor Stanwick.)

Clothing will be put away by remote control

True for those who can afford maid service.

The heating and cooling systems will be built into the furniture and rugs.

Fans are furniture, as are radiators and air-conditioners.

You'll have a home control room - an electronics centre,

Spot on.

...where messages will be recorded when you're away from home. This will play back when you return,...

Not just correct, but conservative, since answering machines turned up way before 2000. (Funny how cellphones arent predicted.)

,...and also give you up-to-the minute world news,

True, radios are part of the electronic center. But didn't they have radios in 1961? Sure they did. So perhaps they were anticipating the fact that local radio stations rarely have real news, but that there would be an internet with lots of news. Impressive!

...and transcribe your latest mail

Transcribe? As in write out what's already written? Did they mean 'read out'? Shelve this under inunderstandable.

You'll have wall-to-wall global TV.

Sure, cable service in different rooms.

an indoor swimming pool

Sure, I have a jacuzzi. (You won't see it if you visit me though, I've lent it to the neighbors. For an undisclosed fee.)

TV-telephones

Webcams, chat rooms, netphoning, etc.

room-to-room TV

Two people in adjacent rooms can run Netmeeting if they really want to.

Press a button and you can change the decor of a room.

Press the emergency 'FIRE' button in a theater and you'll change a lot of decor. Using TV-B-Gone in the dying seconds of a ball game in a crowded sports bar also works.

The status symbol of the year 2000 will be the home computer help, which will help mother tend the children...

"Go to your room and play with your Xbox. Mummy has to teach Daddy how to use a belt."

cook the meals

TV dinners are still with us.

and issue reminders of appointments

My computer beeps whenever it's time for me to go and look at the post-it notes on the fridge for an upcoming appointment.

Cooking will be in solar ovens with microwave controls

Any campfire ignited using a magnifying glass counts as a solar oven. Microwave controls refer to any on-off mechanism, and campfires are either lit or, if I'm responsible for the lighting, not.


Garbage will be refrigerated, and pressed into fertiliser pellets.

No shit.

Food won't be very different from 1961,

Finally, a wrong prediction. We have far greater choice of junk food today.

but there will be a few new dishes - instant bread, sugar made from sawdust, foodless foods (minus nutritional properties), juice powders and synthetic tea and cocoa.

Other than the sawdust, true. Bread is now edible the moment it leaves the oven. Foodless foods are a clear reference to junk food (and celery) so I'll take back that 'wrong prediction' judgement of a few words ago. Juice powders are common, as are synthetic foods. Synthetic doesn't have to mean 'all-synthetic', so tea can be called synthetic even if it has tea leaves, as long as it has fancy chemicals too.


At work, Dad will operate on a 24 hour week.

Typo. They must have meant 'day', not 'week'. PBS Story on Working Family Values, 1998


The office will be air-conditioned with stimulating scents and extra oxygen - to give a physical and psychological lift.

True for those who work in emergency rooms.

Mail and newspapers will be reproduced instantly anywhere in the world by facsimile.

Fax = internet + printer.

There will be machines doing the work of clerks, shorthand writers and translators

Recording equipment has made shorthand writers redundant. And speech recognition and automatic translation tools are commonly used. Note that the prediction didn't say that the machines would actually work well.

Machines will "talk" to each other.

Networks.

It will be the age of press-button transportation.

Elevators.

Rocket belts will increase a man's stride to 30 feet,

True for astronauts doing space walks. (They didn't say that the human doing the striding had to be on earth.)

...and bus-type helicopters will travel along crowded air skyways.

True if you're really rich. And helicopters & planes keep killing birds, indicating that the skies are getting crowded.

There will be moving plastic-covered pavements,

'Plastic-covered' refers to litter. 'Moving' refers to San Francisco, 1989.


..individual hoppicopters

These are personal helicopter packs, which sadly haven't seen much development since their use in the 1940s. Which is a good thing - a drunk hoppicopter pilot is a scary thing to contemplate.

and 200 m.p.h. monorail trains operating in all large cities.

Darn. They had to be more specific. Morons. Anyway, true if you live in Japan.

The family car will be soundless, vibrationless

I recall reading a story in 1995 about BMW having to put vibrating sounds back in cars after tests showed that people driving truly soundless cars were unnerved by the experience... and once hydrogen and/or electric cars are common, cyclists will have to watch out...

and self-propelled thermostatically

Wtf?


The engine will be smaller than a typewriter

Probably true. There have been some pretty large typewriters.

Cars will travel overland on an 18 inch air cushion.

No, hovercars aint here yet.

Railways will have one central dispatcher, who will control a whole nation's traffic.

True in Luxembourg.

Jet trains will be guided by electronic brains.

Ja.

In commercial transportation, there will be travel at 1000 m.p.h. at a penny a mile.

When my company pays, it's even cheaper than that.

Hypersonic passenger planes, using solid fuels, will reach any part of the world in an hour.

Nowhere near here yet. Scramjets will make it possible in 2080.

By the year 2020, five per cent of the world's population will have emigrated into space.

True. These spaces around cities are called slums.

Many will have visited the moon and beyond.

I think a dozen people counts as 'many'. Sure they were Apollo astronauts, but that doesnt make them not people, right?

Our children will learn from TV, recorders and teaching machines.

Ja. Good thing there's no prediction on what they learn from TV.

They will get pills to make them learn faster.

Adderall.

We shall be healthier, too.

Big misprediction.





There will be no common colds, cancer, tooth decay or mental illness.

Ditto.

Medically induced growth of amputated limbs will be possible.

Damn the word 'medically induced', otherwise we'd just say 'lizards' and give them a point.

Rejuvenation will be in the middle stages of research, and people will live, healthily, to 85 or 100.

There exist geriatric marathoners.

There's a lot more besides to make H.G. Wells and George Orwell sound like they're getting left behind....

Think iPods, cellphones, Google.

And this isn't science fiction. It's science fact - futuristic ideas, conceived by imaginative young men, whose crazy-sounding schemes have got the nod from the scientists. It's the way they think the world will live in the next century - if there's any world left!

Well, we should congratulate the world for getting past the 'if'. That probably means thanking Nikita Kruschev for not being overly macho during the Cuban Missile Crisis. (Not JFK, he would have pressed the red button, destroying the world to save it.)

Final score: 43 right, 8 wrong, 2 inunderstandable. Not bad.

Permalink

Friday, January 07, 2005

Friday



Converted the Digital Universe to stereo and told Mark and Brenda about it. Mark will have it ready when he and Randy go to the AAS meeting in San Diego in case the AMNH folks want to use it on the geowall. Also added Brian's additions of the latest 2dF and SDSS surveys, and the exoplanets. Sent a long post telling people about additional features. Brian gave a good response promptly. He's cool :)


Zimbabwe's ranked 151 out of 155 countries in the 2005 Index of Economic Freedom. The only (ranked) countries where the government interferes more with the economy are Libya, Myanmar, and North Korea. What a piece of shit Mugabe is. All of Zimbabwe's a wasteland, and all ZANU PF cadres merely turds.


I didnt realize Estonia was #4 on this list. Wow...

Unsurprisingly, Botswana (Zimbabwe's neighbor) is the highest African country on it, at 37, ahead of Japan and France.

Randy likes the new iron nucleus shower. He suggested we investigate our assumption that all particles in the shower move at the speed of light, since it's pretty clear at an early timestep that the particles have a kinda sphere from the start point.



Had a great discussion with Daisuke about problems with centering his data, what his data meant in spacetime, and finally ended up on pronunciations of Japanese words. That's what you get when a computational linguistics student makes cosmology visualizations.

I suddenly wondered how many cricket teams ever won a match after having to follow on, and Googled (sans quotes) 'winning after follow-on'

And Lo! The Almighty Google had mercy on its faithful client and pointed me promptly to the answer at HowSTAT, a scarily broad site of cricket statistics. Anyway, out of 262 test matches it had on record, 200 were won by the side enforcing the follow-on (unsurprising), 62 were drawn (how many due to unwanted precipation, I wonder) and 3 were won by the side that suffered the follow-on. While such reversal of cricket fortune is rare, I hadnt realized it was *that* rare!



Funnily enough, the side suffering the ignominy of snatching defeat out of the jaws of victory was, in each case, Australia. After their first reversal by the Pommies in 1896 (yeah, eighteen), the next was in 1981, at Headingley (photos). Now I know why the English keep harping on about it! It's not just that they've not had much to celebrate cricketwise since, but also that this is actually something worth harping on about. (Especially if you're a cricket journalist with a pachydermic memory on a slow news day.) Finally, the third instance was twenty years later, when the Aussies were in India (Eden Gardens 2001, to those in the know). A little known piece of fiction resulting from this defeat was that the word Dravidian now has an alternative meaning in Indian English dictionaries, as "one who worships Rahul Dravid".

Went to the geowall show Special Treatment by Applied Interactives in the evening. Interesting, but failed to touch me enough. But it's still a work in progress, so I look forward to seeing it completed.

Oh dear. Helped a dear friend of mine revisit his pain from ages ago, by mentioning a happy memory (us becoming national quiz champs in Zim) from 1994. The wrapper memories weren't that happy. Owww...

Thursday, January 06, 2005

Pleasnowmore

Today was not a normal day. No, it wasn't Chicago getting 7 inches of snow (that only happens between zero and a dozen times a year). It wasn't CNN taking Jon Stewart's advice and canning crossfire. It was me exercising. Ran three miles - on a treadmill. My nose still itches.


Speaking of Crossfire, would CNN have eliminated it if Fox hadn't been kicking CNN's ass when it came to mudslinging debates masquerading as news? Were Jon's comments really what pushed it out, as the CNN CEO suggested, or had this been coming for a while, or both? Was it just because Tschmucker Carlson (right : mug) had been wanting to leave for months and had finally found a new network to pollute? Who cares.... wonder what Jon's next target's going to be... wonder which show's going to invite him next...

    ''Crossfire" began in 1982 and was once a mainstay of CNN's prime time... But as Fox News Channel perfected the format with popular hosts Bill O'Reilly and Sean Hannity, ''Crossfire" lost favor among CNN executives and was moved to the afternoons in 2002. It averages 447,000 viewers each weekday, down 21 percent from the previous season, according to Nielsen Media Research.... -- Boston Globe)


Made two kinds of showers, with primary particle an iron atom (see pic) and an electron. Also modified some code and documentation on the Aires Vis project page.

Hadn't realized the Google SOAP, and other goodies, were available from Matlab. Woohoo!

RegretTheError is funny.

Didnt do any hmsvm today. Did read a few page of Gussenhoven's book. Prosody reading group tomorrow.



Got an email from Daisuke Nagai, a final year cosmology grad student who wanted to know if his data could go on a geowall; I got it up there in a few minutes, and now he's going to generate more data so I can make a stereo animation of dark matter halo stuff. Should be fun! And he's going to use partiview for research, when he saw how quickly and nicely his data could be modified to an easily-visualized format.

Tuesday, January 04, 2005

quick notes

Bought groceries today, including a now-empty box of kolacky. This is probably not good for me, especially since today was a normal day.

I don't get any exercise on normal days.

Tried some variations on the maptask problem, including a hack that added label estimates based on the previous element of a data sequence and a transition matrix. (It failed.) And binarizing count features. (Failed.) And adding a feature that was the number of unique words in an utterance. (Failed.) Well, 'failed' is harsh, since I'm using it in the sense of 'did not improve results'.

I'm not taking into account the fact that the label classes are of different sizes, e.g. we expect to find differing numbers of checks and clarifications and binary questions, etc. Future plan: implement Abe-Zadrozny-Langford method for reducing this weighted multiclass classification problem to binary classification.

Looking at SVMStruct now, specifically the files Thorsten set up to add new SVM struct objects. I so don't want to do this. I need some endorphins. It's too cold to run, and even if it wasn't, I'd find some other excuse.

Discovered SPIDER. Cool! And they need more visualization components. Woohoo, more stuff to do...

Did some complicated googling to find a draft paper that probably wasn't meant to be found. Will resist linking to it.

Discussed showers with Mark. Finally, someone testing my documentation! :) He wants a web interface, so that someone can fill out a form defining the initial cosmic ray / particle and then get back a VRML model of the shower structure. And he suggested replacing points by lines. I wish I remembered properly what went wrong the last time I tried doing that. In principle, it's a good idea, but ... what happened? ... were the lines too small? Trouble is, I can make thick lines in partiview now, so I can actually add this change...

Ah, the temptation to focus on what I can do rather than what I should do... trouble is, whenever I try to do the reverse, I do neither what I can do nor what I should do. [Think : empty box of kolacky.]

Sent off an email to the Physics Education journal asking if they were interested in an article on visualizing cosmic ray showers. They had better be. And to a second GRC Ed/Vis organizer to ask when the poster deadline was. The first organizer I emailed a week ago didn't bother replying.

Had a pleasant surprise from a visualization professor who found my site and invited me to give a talk if I was ever in his neighborhood. And it's a great neighborhood, too! Pity I won't be going there in the near future...

Monday, January 03, 2005

Temperature falling

The temperature is falling here. It's still fairly warm for this time of year, i.e. around freezing. I went for a walk around 10pm, with a single fleece jacket. It was meant to be a short run, but I didnt take any gloves, and couldn't persuade myself to take my hands out of my pocket. Other than that, though, it was a nice walk. Not too many other people walking along Lake Michigan at that time of night. Just a couple of folks walking their dogs.




Applied
for a NASA workshop at the AMNH for visualization; took about an hour to write up my application, peppering it with urls of work I've done. Will have to apply for another ed/vis workshop in the morning.

Why, oh why, don't I do visualization for my thesis? Because I'm a one-trick pony, that's why. Well, not quite. I make the same pony do many tricks.

Speaking of thesis, my efforts to implement a Hidden Markov SVM in SVMLight are going slowly... just read Thorsten's paper on SVMLight, in a book collection of papers, was kicking myself for not having looked for it sooner. I already have HMSVM code, from Yasmin & Thomas, but can't get it to work right. What am I doing wrong?

Met with Gina and Irina in the afternoon, to brief them on what I learnt at NIPS and plan our NLP reading group this quarter. Decided to alternate between Irina's research and mine each week. Meanwhile, will discuss prosody's use in dialog act classification next week.

Wondering if I should really get that suite of dimensionality reduction algorithms together. Not sure how to get it published even if I do get it together. There's a special issue on similarity-based pattern recognition coming up, would be nice to be able to meet that deadline. And absolufinglutely no chance of meeting it, bien sur.

Jo sent me a nice link, to the Killer Quiz, on which I scored just 6/10. Advice to anyone taking the quiz: go with the stereotypes of what a programmer looks like, i.e. long haired if out of academia or post-1970 and suited-up professor otherwise. Do that and you should get about 8.5/10.

Sunday, January 02, 2005

VERITAS

Woohoo! Finally, someone used the collaboration map I made for the VERITAS telescope project... one of the post-docs there, Brian Humensky (who took several stereo photos of the site last year) just sent me an email saying he used it in his presentations to Randy Landsberg's Space Explorers program last week. He has used other stuff we've made, but I thought no-one would ever use the collaboration map, which I think is a cute idea. It's always nice when the number of people using a long-dead project goes from 0 to 1. More than 1 is nicer, but I'm happy with 1 for now.


Speaking of other things Brian used for his presentation: the movie I made of cosmic rays hitting Chicago while working for Randy and the KICP last summer. This is actually rather bittersweet because I so much prefer the interactive partiview models to the movie... aargh! ... well, I may as well accept it. People like movies, so I better make some more. Better dig up those instructions Stuart Levy told me about a month ago. Man, do I so not want to do that... too hard...

Anyway, he also used the Quicktime Panorama that Mark SubbaRao made from the pictures Brian took last year. (Below is the rectangular version of the Mark-Brian panorama.)



Brian's going back to the VERITAS site next week for the next two weeks; we should get lots of new pictures. (I'm pictured out now, but that's another matter. We need to buy Present3d.

Vancouver!

Here's a picture I finally processed from two weeks ago. It was taken in Vancouver, where I was for the NIPS conference. Vancouver's a great place, with superb food, especially East Asian. It rained a lot there, but mostly mild drizzle. Still, my umbrella was useful there. Speaking of which, I wonder in which part of Vancouver I left it, because it sure aint back with me in Chicago.


Yours Truly near downtown Vancouver. I just *had* to take a picture under a sign that says "Welcome to Vancouver, a nuclear-weapons-free zone". Posted by Hello

This picture was taken by Alex Karatzoglou, another student at NIPS with a software demonstration who was staying at the Youth Hostel to cut costs. We had some great sushi here, both low-cost good-quantity hi-quantity, and medium-cost hi-quality lo-quantity. We also went cycling around Stanley Park, where it was soon clear that Alex was a far better cyclist than I was. Especially when I got a flat and had to return to the shop for a new cycle...

Saturday, January 01, 2005

Tsunamis in southern India, a week later...

Ah... the first day of the new year. And my resolution is to blog more regularly. Let's see if that happens.

I stayed home yesterday while fireworks went off. My intention was to get some work done, but of course very little work got done. Still, I happened to be on the net when my parents suddenly turned up, so we talked a while.

My folks live in southern India, about twenty five miles from its southern tip at Kanyakumari, but were unaffected by the tsunami. About three hundred people died at KK beach itself, and a thousand in the province as a whole. Thousands lost their homes. The tsunami struck on a Sunday, when Christian fishermen (a sizeable fraction in this part of India) stayed at home for religious reasons. I don't know if that meant they died with their families (i.e. how close their homes were to the sea) or not. A cousin told me she knew of some Muslim fishermen who had worked that day - rather, who worked that night and then went to nearby inland areas to sell their fish, then rushed back to their homes when the tsunami hit, only to find that their families had been killed.


Surveying the Tamil Nadu coast, Jan 1 2005

Local inlanders have been taking collections for those affected, with little (hopefully) attention to religious boundaries.

As happened in several other places, the sea drew back several hundred meters before the tsunami waves (of which eye-witnesses reported more than one) struck, causing many people to venture into to the sea out of curiosity. This reaction from people was also common, except among those few communities around the Indian Ocean whose oral traditions gave advice on what to do if that ever happened.

Sea gypsies' knowledge saves Thai village: "The elders told us that if the water recedes fast it will reappear in the same quantity in which it disappeared," [said] 65-year-old village chief Sarmao Kathalay...

Anyway, a photograph of the temporary new coast was printed in a local Tamil magazine; my parents didn't know how it was taken, though my dad suggested that it might have been from some cliffs near the coast.

KK beach is a popular tourist resort, and one of the attractions is the Vivekananda Rock Memorial. This is an island just off the shore, with a building and tall walls. Several hundred tourists were on stranded on it when the disaster hit, and were there for several hours sans food and water (though a helicopter dropped some packets later) as the boats that had ferried them to the rock were washed away and local fishermen were understandably too terrified to take them to shore afterwards. (It's not like many local fishermen had boats available in any case.) One story is that the first boat to help out was used by a cop (from out of town?), after which surviving fishermen also helped out. Other reports have the fishermen helping out first. The actual truth is irrelevant, the fact is that the tourists were ferried back to the mainland, mostly by locals who had lost a lot more than the tourists had.

Incidentally, one tourist was a Indian Supreme Court judge.

Rediff.com's Tsunami coverage

Amit Varma has a very well-written piece on several topics, including relief organizations donating too many used clothes that few affected people want to wear, and a very interesting interview with a local official who seems to know what he's doing... "When the disaster occurred, we set up community kitchens and fed them, but we encouraged the affected people to go back to their homes and cook. They did just that. We don't just want to take care of their short-term needs. We need to give them their livelihoods back." The local official had even conducted disaster management courses two months before December 26. Check out India Uncut, Amit's blog, for more information.