I don't have access to the study, but at least FTA is seems like what they did is ask people to come up with a list of soft drink brands after exposing them to one brand. They came up with more alternatives individually than when paired with two others (though it's not clear whether each individual came up with fewer alternatives when in a group, which would be a trivial result, or whether the group came up with fewer alternatives than each of the separate individuals, which would be more interesting). It seems like a pretty big stretch to make general claims about the process of coming up with alternative solutions to problems from this data. In a well-run meeting (I know, they're rare) about a real-world *problem*, it seems like people can bring forth solutions that may be flawed, but that others in the group could help adapt to work. The group really can be smarter than the individuals (and I've seen it happen).
Many posters have brought up the concept of groupthink, which also seems to me to be a separate issue. Groupthink has more to do with people being reluctant to come up with/bring forward alternative solutions when the majority of the group has already settled on a decision.
Essentially I'd say that the issue of problem-solving individually vs. in groups is vastly more complex than can be explored with such a simple experiment. As the article mentions, the study may argue in favor of advertising during events that people watch as a group, but that's about as far as it goes.
The caltech datasets are in my opinion artificial, since they rotate all images in the same direction.
For example, a moterbike always faces to the right, and the 'trilobite' is even rotated out of the plane (leaving a white background) so you only need to estimate the right angle of rotation.
For a less structured set of many pictures of many categories, check out the newer "Caltech-256" dataset on the page I linked - 256 categories, and much less uniform images.
Disclaimer: I work with the MIT algorithms daily and know several of the authors of this work (though I'm not at MIT).
This paper's claim to recognize scenes like the brain does, is overdrawn.
As far as i can tell from their paper (it is a journal version of their cvpr paper) only their low-level Gabor features are similar to what the brain does.
Their low-level Gabor filters are indeed similar to V1 simple cells. The similarity between their model and the brain goes a lot further, though. The processing goes through alternate stages of enhanced feature selectivity with roughly Gaussian tuning (the S layers) and pooling over spatial location and scale via a max operation (the C layers). If you read more papers from their lab, there is a significant amount of biological plausibility in both of these operations, and a great deal of effort has gone into tuning the various layers to behave in accordance with physiological data.
The rest of the paper uses the currently popular bag-of-features model, which is a model that discards all spatial information between image features, which i don't think the brain does.
The model is roughly equivalent to a bag-of-features, but with the nice feature (from a biologist's perspective) that it builds the bag in a biologically plausible way. The features themselves are picked randomly from natural images in a training stage that takes the place of human development. Discarding spatial information makes the model a lot more tractable, and it isn't clear what role spatial information plays in the processing of the ventral visual system, which is what their algorithm models.
Furthermore, for classification algorithms they consider a Support Vector Machine and Boosting. Both of these classifiers are certainly not comparable to what the brain does. Why not use a neural network if they aim is to mimic the brain?
They use these classifiers on top of their algorithm simply to determine how good the model was at extracting relevant feature information. Since they want to quantify how much information is there, it is wise to choose the best method they can to locate the information.
Furhtermore, they only conside feed-forward information, where research shows that there is at least as much information going back as there is going forward.
Feedback is definitely very important (this is what my own research is about), but feedforward accomplishes a lot with a vastly simpler computational model.
Don't get me wrong, it is still a nice paper, with good results.
(however, all Caltech datasets are highly artificial, with objects artificially rotated in 1 direction)
So, nice paper, but to compare it with the workings of the human brain is too much.
Here are the Caltech datasets they used: vision.caltech.edu. I think the "artificial" datasets you refer to are the "3D objects on turntable," which are a bit artificial. However, the images they refer to in the paper discussed here are from the Caltech-101 dataset, which consists of real-world images of objects from 101 different categories - most of the images are not at all artificial.
Because of this, many people in my department (experimental psychology) have started turning our manuscripts into PDFs and posting them on the web once they're accepted for publication. In that way, we preserve the peer review system and the journal system while simultaneously giving access to those who can't readily acquire the journal. Yes, we know it's not legal, seeing as we signed over all the rights to the publisher, but we feel it's morally appropriate to let scientific knowledge be free.
Read the fine print of the copyright agreement carefully. Oftentimes this *is* actually legal, as long as you post the final revision rather than the pdf that actually appears in the journal. They may also require some disclaimer stating that it isn't exactly the published version, with a url to the paid, published version (YMMV).
It's still impossible to get a Wii at MSRP (i.e. you have to buy from eBay scalpers), and I don't see this getting any better before Christmas...
I just got mine off of Amazon a few days ago. I clicked the little "X used and new from $600.00" thing just to see who was selling them, and the top thing on the list was a Wii direct from Amazon for regular price. I quick hit "Add to cart" before it could vanish and nabbed it. Keep trying - they're coming in regularly now.
If it uses Bluetooth as it is supposed to, what is to stop the Wii remote being used on a PC or even a PS3 if you wanted to? What's the point you may ask - well it would make for useful mouse replacement for presentations, or just for couch surfing.
Apparently it's trivial to at least use it as a gamepad: link. Pointer functionality can't be far off...
I worked on one of these projects when it was just getting started 5 years ago, and this was one of the problems we had to look in to. The CubeSats were destined for Low Earth Orbit (~300 km was our design case), and would deorbit due to aerodynamic drag in a month or two.
I seems that one way to encourage new discoveries is to learn how to cultivate or induce a state of mind or being that will make oneself more receptive to tangential thinking - by that I mean that moment where one takes a step back and "the light comes on" about something completely unrelated to the current course of research or study. This, IMHO, would be be open-mindedness, or egolessness. Too bad a massive ego is a prerequisite for tenured college professorship - I guess they won't be teaching how to do it.
You're absolutely right about egolessness, but not quite right about college professors. I had the privelage of doing my Ph.D. work at one of the world's greatest engineering institutions, and "egoless" and "open-minded" is an excellent way to describe the best researchers there (student and professor). They are highly confident, to be sure, but open to new views and being challenged. It's generally only the mediocre (and thus insecure) that have to hide behind titanic egos.
Try giving out a card with just your name and number, and see how many ask for your email address as well.
I loved Danny Ocean's business card in "Ocean's 11" (the new one) - plain white card, just said "Daniel Ocean". If you needed to know anything else, it could be written on the back.
I know that several brazilian ISPs have asked for increased bandwidth from our biggest (perhaps the only) backbone and had it denied, as it was not possible.
Maybe if they hadn't all asked at once they could've worked something out. There's no way we can increase bandwidth to several brazilian ISPs all at once!
In order to use the p0rn site he ran, you had to either pay money or spend time identifying captchas.
I saw a talk recently by Luis von Ahn, one of the inventors of the captchas. There were two interesting ways he said people were getting around captchas. One was a real-time approach similar to what you describe. Rather than storing a big database of these things, the bot that was signing up for email addresses or whatever would, upon encountering the captcha, sent that image off to someone browing the porn site (posing as a legitimate captcha - "We need to verify you're a person and not some bot stealing our porn for another site"). In order to continue browsing, the user would have to solve the captcha. Naturally they tend to do this very quickly and accurately:)
The second approach was simply to set up captcha solving sweatshops somewhere in Asia with cheap labor, with people paid a few cents an hour to sit and solve captchas all day. This brought the cost of a new email address up to something like 1/3 cent, which for many spammers is still a viable price. The cost does limit this approach, though, so the captcha still helps.
The interesting thing about both of these strategies is that they use humans to solve a problem that is difficult for computers, which is von Ahn's research area - he's also one of those behind The ESP Game (caution - this can be shockingly addictive). There's essentially nothing that can be done to defeat either approach without also making a system a huge pain in the ass for legitimate users. From this point of view, spending time trying to come up with more advanced captchas is kind of pointless.
Pursuant to the introduction of this new six-axis sensing system, the vibration feature that is currently available on DUALSHOCK® and DUALSHOCK®2 controllers for PlayStation and PlayStation®2, will be removed from the new PS3 controller as vibration itself interferes with information detected by the sensor.
Apparantly the good folks at Sony have never heard of low-pass filtering. I'm pretty sure the rumble pack puts a disturbance into the sensor at a MUCH higher frequency than the player-caused motions it needs to pick up. It really shouldn't be a problem to have both.
I call this the principle of "Think, THEN do." I occasionally mentor groups of engineering students, and this is one of the hardest lessons to convey. Many experienced engineers, even "good" ones, struggle with this.
Google's ad placement algorithms are getting better and better. The page with the video had "Top New Beetle Prices" and "San Francisco Dentist" - I guess they know what the likely result of operating this monster is!
It is likely that DARPA has become incapable of inovation because of internal politics so they need to attract new ideas from the out side.
DARPA is a funding agency, not a research institution. They actually have very few employees, who are mostly there to identify promising research areas and allocate money to invest in them. The actual research is done by academic and industrial research groups. Incidentally, many if not all of the DARPA project managers are actually very good research scientists and engineers who take 1-3 years away from their normal work to work for DARPA, not career bureaucrats (wow, that's a hard word to spell) mired in politics. Most of them really view it as an important public service.
As a tax paying citizen of the United States, it sure is frustrating to try to find the results of DARPA research.
Yes, they do research in defense but shouldn't there be a little more than a tiny graphic or blurb about what work they're doing? Couldn't they at least take the time to write an abstract or 1-2 page paper with unclassified information on each project?
Try instead going to Google Scholar or another academic index and searching on the titles for various DARPA projects. Having worked on several DARPA-funded projects, I can tell you that there is generally a significant emphasis on publishing results. DARPA-sponsored work probably results in dozens or hundreds of articles in scientific journals a year, all of which are available to the public.
When you say your alma mater "has produced better papers in these fields" you should have a look at the acknowledgements section of these papers. Chances are pretty good many of them will have a statement like "This work funded in part by DARPA (or NSF, etc) grant number XXX."
I'll be impressed with no crashing into each other, before they worry about compliance with all traffic laws. How will the robots recognize the speed limit in their area, or will they all crawl along at 10 mph, impeding the flow of traffic?
Um, that's why they call it a "challenge" - because they don't know how to solve the problems yet. If it weren't really hard there would be no need to do this sort of contest.
One thing is for sure: neuroscience is making it more difficult for a spirit to hide in our mushy insides.
For an excellent introduction (at the scientifically-minded nonexpert level) to what neuroscience is doing in this area, check out the book "The Astonishing Hypothesis" by Francis Crick (yep, the same Crick as in DNA - he spent the latter half of his life working on consciousness). The "Hypothesis" in question is that everything that we are arises out of the activity of the "mush" (as you call it) in our brains. wikipedia article
For a more technical overview of the same issues, try "The Quest for Consciousness" by Crick's longtime collaborator Christof Koch. More info at www.questforconsciousness.com.
Does anyone else think it's funny that in racing (say, at the Olympics) when someone crashes or otherwise can't complete the course it's a "DNF" - for "Did Not Finish"?
I wonder how dificult it would be to send a robotic mission that not only lifts the orbit but actualy send the hubble into space to orbit the sun outside our own planets gravity.
There are lots of considerations that make this infeasible, aside from the pure challenge of a robotic mission of this complexity. A big one is that Hubble is in Low Earth Orbit (LEO), at an altitude of roughly 600 km. This means that one side of it is facing a nice, warm Earth, which helps to keep the temperatures on board in a comfortable range. If you now move the thing out to solar orbit (no small feat), it would be facing cold dark space on all sides. I'm sure it doesn't have the power collection and heating capability to deal with the ensuing temperature drop.
I'm wondering exaclty how different the images could be if they weren't effect by earths gravity or if the hubble was actualy closer.
Um, not different at all. The Earth's gravity has no effect on the images Hubble collects (at least at anything near the scale resolvable by the telescope). Same story with distance - most of what Hubble looks at is so exceedingly far away that Hubble could be at Pluto without making any noticable difference.
Light reflected from Earth could be a minor problem, but I think in the visible spectrum this isn't significant either. For an IR telescope like Spitzer it's important to get away from Earth to get decent images, so they put it in a heliocentric orbit similar to Earth's.
Slashdot Mirror
Many posters have brought up the concept of groupthink, which also seems to me to be a separate issue. Groupthink has more to do with people being reluctant to come up with/bring forward alternative solutions when the majority of the group has already settled on a decision.
Essentially I'd say that the issue of problem-solving individually vs. in groups is vastly more complex than can be explored with such a simple experiment. As the article mentions, the study may argue in favor of advertising during events that people watch as a group, but that's about as far as it goes.
I agree with the rest of your post, though.
Oh, and it rules.
Clearly you didn't see this comment.
I worked on one of these projects when it was just getting started 5 years ago, and this was one of the problems we had to look in to. The CubeSats were destined for Low Earth Orbit (~300 km was our design case), and would deorbit due to aerodynamic drag in a month or two.
Wait, I forget - just how many is a brazilian?
The second approach was simply to set up captcha solving sweatshops somewhere in Asia with cheap labor, with people paid a few cents an hour to sit and solve captchas all day. This brought the cost of a new email address up to something like 1/3 cent, which for many spammers is still a viable price. The cost does limit this approach, though, so the captcha still helps.
The interesting thing about both of these strategies is that they use humans to solve a problem that is difficult for computers, which is von Ahn's research area - he's also one of those behind The ESP Game (caution - this can be shockingly addictive). There's essentially nothing that can be done to defeat either approach without also making a system a huge pain in the ass for legitimate users. From this point of view, spending time trying to come up with more advanced captchas is kind of pointless.
Google's ad placement algorithms are getting better and better. The page with the video had "Top New Beetle Prices" and "San Francisco Dentist" - I guess they know what the likely result of operating this monster is!
When you say your alma mater "has produced better papers in these fields" you should have a look at the acknowledgements section of these papers. Chances are pretty good many of them will have a statement like "This work funded in part by DARPA (or NSF, etc) grant number XXX."
For a more technical overview of the same issues, try "The Quest for Consciousness" by Crick's longtime collaborator Christof Koch. More info at www.questforconsciousness.com.
Does anyone else think it's funny that in racing (say, at the Olympics) when someone crashes or otherwise can't complete the course it's a "DNF" - for "Did Not Finish"?
Light reflected from Earth could be a minor problem, but I think in the visible spectrum this isn't significant either. For an IR telescope like Spitzer it's important to get away from Earth to get decent images, so they put it in a heliocentric orbit similar to Earth's.