Interactive Machine Learning

Don't let the title of this talk (Interactive Machine Learning) mislead you. As Dan Olsen says, "I am not a Machine Learning guy, so if you say Laplace in front of me, I will look at you with a blank stare". What the talk is about is HCI.

"[HCI] - why is it interesting?.. Every time I think about computer science and I think about the future I always think about Moore's Law. Mainly because as I tell my students: if you choose a problem for which Moore's Law will solve it, why are you wasting your time. If we've got exponential growth in computer speed; if we've got exponential growth in memory, we've got lot's of available data. The one thing that isn't growing is human capacity. The ability to concentrate, the ability to see, the ability to touch, the size of our fingers, those things are not changing."

"Nobody wants to become the borg. People want to live the lives they live and they want to be assisted by their computing technology, they do not want to be dominated by it."

-Dan Olson (Interactive Machine Learning)

Good design principles from...

Image Processing with Crayons, a drawing application where you don't have to stay within the lines:

- Painting is the fundamental metaphor.
- Users can become semi-proficient in ~ 4 minutes.
- ML algorithm must train in less than 10 seconds, preferably 2.


How long can I neglect my robot, an experiment to teach a robot what is safe and what is not safe:

Digital input device a:
- You look at where you want to go and you go there.
- A great advance over the joystiq?

Digital input device b:
- Scribble what is safe in blue and what is unsafe in red.
- You can teach anyone how to do this in ~ a minute and a half.


"It cannot take days to train, it must take seconds. Otherwise you don't have an interactive loop." There is a user with some kind of user interface (e.g., painting with crayons - drawing on transparent layers) who labels some artifacts (e.g., pixels). Convert that into features, train a training algorithm, apply it to a bunch of unlabeled artifacts and provide feedback to the user. This the loop. Focus on it.


Driving robots are being funded by the military?

Apple nano Money

Finding checks in the mail is always fun, but if I knew it would take this long to get my Apple iPod nano scratch settlement money, I would have asked for $40.00 off the purchase of a new Apple iPad.

The End of Anonymity, The Beginning of Privacy

Contrary to what the movie Fight Club may have you believe, you actually are a unique snowflake. For evidence of this, just take a look at the Netflix Prize dataset (it contains 500 thousand records - users, and about 200 movie ratings per record). Statistically speaking, 90% of the records do not have a single other record that is more that 30% similar to it. In other words, the vast majority of Netflix users have rated a very unique set of movies.

Netflix, being very concerned about the privacy of their users, has removed or anonymized so called personally identifying information (name, email, age, address, etc) before releasing the dataset. However, the very notion of personally identifying information is flawed. In fact, all information can be personally identifiable. That is, any information could be used to identify an individual.

Taking a look at the Netflix dataset again, we find that on average, two movies is enough to reduce the candidate records to eight. And four movies is all it takes to uniquely identify one record. Another way to put it is, if you know just four movies that your friend has rated and you know that he is in the Netflix dataset, then it's very likely that you can find his record and learn the other movies that he has rated. This could be potentially embarrassing (gay porn) or dangerous (politically charged movies) for him. In fact, people have carried out attacks like this by linking the "anonymized" Netflix data with publicly available IMDB rating data to learn the identities of several "anonymized" Netflix users.

Previous definitions of privacy that were based on personally identifying information (quasi-identifiers) were flawed for this reason. k-Anonymity (syntactically transforming the dataset so that quasi-identifiers must appear in at least k records) does not guarantee privacy. Privacy is not a property of the data, but a property of computation carried out on the data. A better definition of privacy is differential privacy. Differential privacy basically means that including or not including a particular record has no significant effect on the computation result. Or in other words, your privacy has the same chance of being violated whether you participate in the computation or not.

[Reference Video]

Computational Developmental Robotics

[Comic: Toothpaste For Dinner]

At first glance, human babies don’t seem that bright. They eat, they sleep, they cry. But they also learn at an astounding rate. While they are still in the womb, they learn proprioception, where their limbs are in relation to themselves, and how to move them. When they are born, they need to quickly learn how to breathe (just-in-time learning, if you will). They also get access to an entirely new sense - vision. Babies learn face recognition, structure from motion, and depth from stereo faster and better than the state of the art computer vision algorithms. Similar things can be said for a baby learning how to move about and manipulate its environment (robotics), learning how to understand and communicate with others (computational linguistics), and learning how to enjoy music (beat detection). Even more incredible, is that all this learning is going on while the baby’s brain grows from a single cell when the baby is in the womb, to the complex network of neurons in a developed brain. It’s like learning how to fly a plane as it is being assembled from parts all around you.

Artificial intelligence is a field with a lot of cool stuff happening. We have computer programs that can beat human grandmasters at chess, spam email filters that operate at over 99% accuracy, and programs that can compose piano music indistinguishable from the greats. But, how much of this can we actually call artificial intelligence, and how much is hand-wavy fakery? Can Deep Blue learn how to play Go? Can a spam filter learn to love? Why is it that a baby can learn to solve any problem and adapt to any environment, while we haven’t even come close to creating a robot with such versatility and adaptability?

The problem may be that artificial intelligence has been too focused on advanced behaviors like game playing (chess), classification (spam filter), and rule breaking (computational creativity) and less so on fundamentals like motivation, environment modeling, and agent self-introspection. After all, you must learn how to crawl before you can run.

tl;dr – Babies are awesome. We need to learn how they learn.

Don't Pepper Spray me Brotender!

This is what happened last night. My friend David and I went to Trinity nightclub in downtown Seattle at around 9:45pm (Wednesday night is free cover and $3 beers). We go in and the bartender immediately checks our IDs. The bartender notices I have a half empty plastic bottle of Cha Dao Black Tea and Coffee Yin Yang (awesome drink by the way), and asks me to hand it over. I don't really understand why, but I reluctantly do so. Then I jokingly say to David, "I guess they have a no tolerance policy here". The bartender overhears this and gets really angry. He basically says if we don't like it we can leave. I feel the bartender is being a dick, so I ask to see his manager. This seems to piss him off even more. He says that it's not possible and then tells us to get out.

At this point, I feel the bartender is being a total douchebag. I go with David past the bar area and walk to the dance floor area looking for someone less dickish that I can talk to. To my utter surprise, the bartender has followed me and he's carrying a bottle of pepper spray and is threatening to use it on me. I really don't want to get pepper sprayed so David and I quickly leave.

I realize that you're only getting my side of the story here, but I just want to highlight two things that can be confirmed by the other patrons and David:

1. The bartender chased after me with a bottle of pepper spray in his hand and threatened to use it on me.
2. I never made any threats, or even hinted at violence or getting physical. I made it clear to the bartender that violence was the last thing I wanted.
   

tl;dr - Bartender at Trinity goes on a power trip and threatens to pepper spray me. I never once indicated a desire for violence or physical confrontation.

On a lighter note, I find it funny that I seem to be prone to bad luck whenever I go out on my birthday.

4 Things the Post Office Can't Do

The post office recently failed to deliver a package to me. In trying to understand why, I've discovered 4 things that the post office simply cannot do.

1. Deliver my package.

This is the online tracking page for the package in question. Notice how it made it all the way to Bothell, WA (where I live), but it was never actually delivered to me. Instead it was marked "Return to Sender" without my knowledge.


2. Explain why they can't deliver my package.

So I called my local post office looking for an explanation. After getting tossed around their phone tree a few times, the response I got was: generally, a package getting marked "return to sender" means there's something wrong with the address.

3. Help me figure out what went wrong.

That's alright I think, mistakes happen. However, when I then ask what the incorrect address was that caused the package to be marked "return to sender", nobody can answer me. I'm beginning to smell some BS. I also checked with the seller later, and they confirmed that they shipped it with the correct address.

4. Fix the mistake.

I don't really care whose fault it is, I just want my package delivered. So I ask them to correct the address and deliver it to me. The package is literally in the same town as me, so I figure this can't be that hard of a request. It turns out this is impossible. After getting escalated three times, the response I got was along the lines of: our system is not capable of doing such a thing.

Well, thanks post office, I hope you're happy. The sender is now sending me the package again, this time via Fedex.

Things That Don't Exist in Reality, Suck

I've been searching for a certain poster from the movie "Cloudy with a Chance of Meatballs". It's the one with Nikola Tesla being a badass rockstar. The artist, Pete Oswald, has kindly provided a digital version on his blog.



But I really want to buy a real-life version to put on my wall. So far, the closest thing I could find is in the top-right corner on the back of the Cloudy Artbook.



Update: I'm not the only one that wants it apparently.

Droid Navigation

I sold my old iPhone and got a Droid with car dock from Motorola. Google Maps Navigation on the Droid is well designed. The car dock is not.

The dock holds your phone - and that's it. If you want power, you need your own usb car charger. If you want audio from the phone to the car stereo, you need another cable. One nice feature though, is that it allows for easy rotation of the phone between landscape (useful for navigation / watching YouTube videos) and portrait (useful for calling people / web browsing) orientations.

Google Social Circle

Normally when I do a vanity search, I use my full name. Today I did a search for just "Andy" for fun. I wasn't expecting this:


At first I though Google Search was broken. How could this story about me be on the first page of search results for "Andy"? Then I realized I was looking at a new feature, called Google Social Circle.

It's quite ingenious actually. While Facebook and Twitter are first building a social network and then building (or in some cases hacking) search on top of it, Google is approaching the problem of social search from the opposite direction. They've built a rock solid search engine, and now they're experimenting with the social side of search. I think it boils down to this: people are going to be more likely to visit a link that is recommended, written by, or is in some way associated with one of their friends.

Disclaimer: I work at Google (on unrelated projects).

Depressing Chart

Source: [OkCupid Blog]

Time-Lapse IKEA

I've watched my fair share of time-lapse IKEA videos on YouTube, but let me tell you, this one makes all those others look normal-speed.

And the reveal...

6 Things to Do in Australia

1. Watch the New Years Eve fireworks in Sydney.

2. Learn how to play cricket.
(and then lose to my sister)

3. Try boogie boarding at Bondi Beach.


4. Have some kangaroo on the barbie.

5. Buy "CANDY posing pouch".

6. Go to jail. :(

But not necessarily in that order...

Disappearing Christmas

When I say Christmas can disappear, I'm not talking in the metaphorical sense, like your parents forgot to buy the Christmas presents. I mean that it is possible to experience a shorter version of it, or in the extreme case, no Christmas at all, due to travel around the Earth.


Let's look at an example. Flying from Los Angeles to Sydney takes about 15 hours. A flight leaves at 6pm Christmas Eve. In 10 hours it crosses the international date line. Just before the plane crossed the line, the time was 4am Christmas Day in Los Angeles (GMT -8), however for the passengers on the plane, the time was 12am (midnight) Christmas Day, since they were at GMT -12, just to the east of the international date line. When they crossed the line, time jumped ahead by 24 hours and it is now 12am December 26. Continuing to travel west, the plane will arrive in Sydney at 3am December 26 local time (GMT +10).

As you can see, the passengers on the plane have no experience for any time between midnight Christmas Day to midnight December 26.

Computer Games

When computers finally managed to beat us at chess, a game of ingenious human invention, we did not say they were now officially smarter than us. What we did was give computers another game - another puzzle to struggle with.

Go is a board game that is a magnitude more complex than chess. Whereas computers could memorize well studied opening books for chess, such a strategy is unmanageable for Go, since there are 361 possibilities... for the first move alone. Today's fastest computers with the best Go algorithms are still beaten by human players with a few months of experience - mere novices.


I believe that computers will one day manage to beat us at Go. But that is a discussion for another time. Now I want to talk about what game, what puzzle we will toss future-computers after that, to assuage ourselves that NO - they are still not smarter.

Settlers of Catan is a board game that is a magnitude more complex than Go. Whereas future-computers only had to think about winning or losing for Go, a binary 1/0 that they are familiar with, such a strategy simply doesn't work for Settlers, since bartering between players involves the hazy variables of human psychology that computers are ill-equipped to define. First there is the game. It is played on a board, which has over a million different possible configurations - and that's before any player has had a turn.

Then there is the meta-game. It involves players bartering with each other, forming alliances, and knowing when to stab their former partners in the back. To give you an example of how complex the thought process can be, we need to take a quick diversion.


From personal experience, I've seen a human player propose an ingenious trade, that would have given her an enormous advantage. To fully understand the significance however, you first need to understand more about the game. Victory in Settlers is achieved when a player has 10 victory points. There is a longest road development card, which is awarded to whichever player has the longest road* of at least 5 segments. It is worth 2 victory points.

You also need to understand more about this particular instance of the game. There were four players. The human player and another player (let's call him the opponent) were both just mere points away from victory. One of the two other players currently held the longest road card. However the opponent had two large sections of road, and if he was able to connect them, held a good chance of gaining the longest road card for himself.

The human player proposed that she refrain from building a road at a position where it would block the opponent from connecting his two road sections in return for him giving one of his resource cards on every turn to her for the rest of the game (like a tax). Right as the opponent was about to agree to this outrageous deal (remember he still might not have gotten the longest road card), the human player realized it was better to take away all possibility of the opponent gaining the longest road card, and she built the road anyway.

What will happen when future-computers beat us at Settlers of Catan? That's when we can say they are officially smarter than us - and be right.


* The length of a road is the longest possible segment, not counting branches.

Keyboard Cat Piano

What better way to test a new piano?

P.S. - I'm testing my new HD camera at the same time. :)

Running the Seattle Marathon

About three months ago, my friend David and I decided to run a marathon. Neither of us had run so much as a 10k before. Although it was a spontaneous decision, we took it seriously. I found a beginner's training schedule and we stuck with it as best we could. Rain or shine, night or day, oncall or not, we ran.

Marymoore park was our preferred training location. We would always start at the park, run out half our training distance, turn around, and try to make it back to the car. I had blisters on my feet, sore joints, and aching legs, but we would always made it back. We ran that trail so many times, that it became a part of me; I learned every twist and turn, and every mile marker. We could run it with our eyes closed, and sometimes we did, in the pitch black of night.

The night before the marathon, my roommate cooked. I went to sleep with a belly full of pasta. I was woken up at 4am by an unexpected text message. "What are ya doing man?" it read, but I was too bleary eyed and preoccupied with the marathon to reply. I needed to get going anyway. When we hit traffic driving to Seattle Center, David kept busy by measuring the rate of raindrops hitting the windsheild. "It's slowing down," he told me.

We managed to park and walk to the race start with a few minutes to spare. My Dad had come to support me and take pictures. It was the first time I had seen him in a month. He didn't say anything about my hair. The raining stopped just as we came to the start time.

The first half of the marathon flew by for me. Seeing everyone else running with me, and hearing the spectators cheer us on really got me going. At that moment, I felt like I could run 100 marathons. We started near the back, but we kept passing runners. As we ran across the I-90 bridge, we caught up with the 4 hour 45 minute pace runners. We managed to stick with them through the halfway point, giving us a split time of 2 hours 22 minutes.

At mile 15, things got a lot harder. Pain and fatigue caught up with me and I couldn't keep up my pace. I had to slow to a walk, and the aching that was masked by endorphins really kicked in. It was the worst I felt during the entire race, and perhaps my entire life. I wondered how I could ever finish the remaining 11 miles. Giving up was so tempting. But I remembered the guy at mile 9, holding up the sign that read "Pain is temporary, pride is forever". I remembered the smile of the little girl at mile 12, cheering and making more noise with her little cowbell than anyone else. I remembered the man with the loudspeaker at mile 7, saying how he was so proud of all us. I remembered all the reasons that I had to run. And I kept going.

Each mile after that was a struggle. The time between mile markers grew and felt like an eternity. At mile 20, my Dad called me, wondering how I was doing. At the time, my phone felt like a brick in my hand. He asked if I wanted him to come pick me up. "No", I said without hesitation. I had made up my mind. There was no way the day was going to end without me finishing what I had started.

Just short of mile 25 I heard David yell out in pain. I turned around and saw him holding his leg. It had cramped up. David told me to keep going without him, but I wouldn't have it. I wasn't about to abandon my friend. Not when we had gone so far, and were so close to the finish. I waited for David to recover. Then we limped/walked/ran/sprinted the final mile.

It took me 5 hours 42 minutes and 10 seconds to run my first marathon. So why did I do this? Why did I choose to put myself through such pain? For me, it's more than just being able to say I've run a marathon. It's about pushing the limits of my ability, experiencing new things, and most importantly, achieving something that I really want.

420 Possibilities

1 black 4x2 bookshelf from Ikea + 2 blue bins + 2 red bins = 420 possible arrangements. What you see above is the one that defines me as a person.

Chef

I found out that being a chef has its perks. And I'm not just talking about scaring little kids with your big bloody knife. I went to a busy Indian restaurant with some friends after work, and was still wearing my chef's outfit (though I had put away the knife). I had a feeling our service was a bit faster than usual. We were seated immediately while others had to wait. Our plate of rice and glasses of water were always refilled quickly. Our food was very good. Not only that, but after the main course, we were presented with complimentary tea and desserts. Right before we were about to leave, our waiter asked me what I had thought of the food. He looked disappointed when I told him I wasn't really a chef...

You

I never thought a cellphone comercial would make me want to cry...

Worst Song or Best Song?

Scientists have carefully constructed the most unwanted music. However, in a weird reverse-Frankenstein, the 22 minute song is actually pretty good. The same auditory elements that are so undesirable, when combined together, actually make for a pretty awesome song. And yes, there is opera rap.