Futurismic

A computer scientist at the University of Pennsylvania has decided to mimic the way children learn to recognise faces and voices in order to speed up the artificial learning curve of intelligent systems:

Using novel learning algorithms that combine audio, video, and text streams, Taskar and his research team are teaching computers to recognize faces and voices in videos. Their system recognizes when someone in the video or audio mentions a name, whether he or she is talking about himself or herself, or whether he or she is talking about someone in the third person. It then maps that correspondence between names and faces and names and voices.

“An intelligent system needs to understand more than just visual input, and more than just language input or audio or speech. It needs to integrate everything in order to really make any progress,” Taskar says.

The information Taskar’s team feeds into the system is free training data harvested from the Internet. Attempts to teach computers visual recognition in the pre-Internet age were hampered in large part by a lack of training content. Today, Taskar says, the Internet provides a “massive digitization of knowledge.” People post videos, comments, blogs, music, and critiques about their favorite things and interests.

Hah! And they said YouTube would never do any real good! Taskar’s computer seems destined for a life of increasing frustration with irresolvable plot lines, though, as they’re training it by showing it episodes of Lost:

As Tasker’s team feeds more data about Lost into the computer—such as video clips, scripts, or blogs—the system improves at identifying people in the video. If, for example, a clip contains footage of characters Kate and Anna Lucia, after being taught, the computer will recognize their faces.

“The alogorithm is learning this from what people say, or from screenplays as well,” Taskar adds. “The screenplay doesn’t tell you who is who, but it tells you there’s a scene with [two characters] talking to each other.”

Taskar says the information the research has produced can be helpful in many ways, particularly in searching videos for content. Currently, if a father is searching for a photo of his daughter playing with the family dog in his gigabytes of photos and videos on his hard drive, unless the photo is tagged “daughter playing with dog,” chances are he isn’t going to be able to find it.

Well, that’s your consumer-level pitch, sure, but the system will be too large and ungainly (and expensive) for Joe Average for a long time. Tasker should probably talk to the UK government… that panoply of CCTV cameras keeps growing, and it costs big money to hire people to watch their output. And what could possibly go wrong with putting an automated recognition system in charge of crime prevention? [image by bixentro]

Younger readers (or those with spousal units prone to nagging about excessive time spent in front of a computer) may wish to arm themselves with the news that internet use appears to restore and strengthen brain function, particularly in the elderly. In other words, surfing the web is keeping your brain young and fit. [via The End Of Cyberspace; image by mhofstrand]

For the research, 24 neurologically normal adults, aged 55 to 78, were asked to surf the Internet while hooked up to an MRI machine. Before the study began, half the participants had used the Internet daily, and the other half had little experience with it.

After an initial MRI scan, the participants were instructed to do Internet searches for an hour on each of seven days in the next two weeks. They then returned to the clinic for more brain scans.

“At baseline, those with prior Internet experience showed a much greater extent of brain activation,” Small said.

After at-home practice, however, those who had just been introduced to the Internet were catching up to those who were old hands, the study found.

Of course, this doesn’t take into account the theory that the structure of the web means we only ever get exposed to ideas that we already find agreeable, but I remain unconvinced of that notion, anyway. A brief glance at history shows that people were always perfectly capable of ignoring information that they found unpalatable, long before the internet (or even the printing press) existed…

But while you’re advising grandma to fire up Firefox, be sure to remind her that it’s an aid to learning, not a replacement for it. Recent research shows that we learn much more quickly if we take a chance to answer incorrectly before looking up the correct response… so try guessing before you Google it, in other words.

Robots have been mobile for decades, but they’ve only ever been able to go places for which they had a map or set of directions stored. That’s all changed thanks to a team of roboticists from Munich, who’ve built the first robot that can be unleashed into unfamiliar territory without a map. How does it complete its journey? It asks for directions, of course:

ACE uses cameras and software to detect humans nearby, based on their motion and upright posture. As it closes in on a likely helper, ACE’s “head” – bearing a touchscreen and a second screen displaying an animated mouth – turns to face the chosen person.

A speaker working in sync with the animated mouth is used to get the person’s attention and to ask them to touch the screen if they want to help. Willing guides are then asked to point the robot in the correct direction, with the response being analysed by posture recognition software. Direction set, ACE says “thank you” before trundling off.

Pointing, rather than telling the robot where to go, avoids confusion caused by the fact that the robot and the facing pedestrian each have a different sense of left and right.

Although it interacted with 38 people over a period of nearly five hours – ACE did eventually reach its destination. In fact, the team report that the robot was making very good progress until it reached a busy pedestrian area where its own popularity became a problem.

The current rarity of mobile robots in public spaces is obviously a big factor here; in a few more decades, we may barge past lost robots on the pavement as quickly and guiltily as we do homeless people or street-drunks.

The principle on display here is that of robot-human interaction in order to gather environmental data to complete a task or journey, which is all well and good, but it’s a proof-of-concept more than anything else. If all you needed was a robot that could navigate an unfamiliar cityscape, it’d be far easier to kit it out with good visual sensors and a GPS unit.

Hell knows this would be useless for military applications; if your super-killbot had to stop at every enemy checkpoint to ask the way to headquarters, I dare say the best place it would end up would be a long long way from anything at all… [story via regular commenter Evil Rocks; apologies to Paul McAuley for the headline]

Unless they start to adapt quickly, colleges and universities could become irrelevant in little more than a decade. So claims Professor David Wiley, at any rate, using arguments that should be familiar to die-hard internet denizens and futurists:

America’s colleges and universities, says Wiley, have been acting as if what they offer — access to educational materials, a venue for socializing, the awarding of a credential — can’t be obtained anywhere else. By and large, campus-based universities haven’t been innovative, he says, because they’ve been a monopoly.

But Google, Facebook, free online access to university lectures, after-hours institutions such as the University of Phoenix, and virtual institutions such as Western Governors University have changed that. Many of today’s students, he says, aren’t satisfied with the old model that expects them to go to a lecture hall at a prescribed time and sit still while a professor talks for an hour.

Higher education doesn’t reflect the life that students are living, he says. In that life, information is available on demand, files are shared, and the world is mobile and connected. Today’s colleges, on the other hand, are typically “tethered, isolated, generic, and closed,” he says.

It’s the “open everything” argument, of course, but it’s given a certain extra weight in this instance because Wiley lectures at Brigham Young University, a small private university owned by the Mormon church; if they can see the writing on the wall and admit to it, then change is definitely afoot (although Wiley makes the point that establishments like Brigham Young offer “a religious education and the chance to meet and marry an LDS Church member”, which is effectively a kind of social network attraction, albeit a non-technological one). [via Technovelgy; image by Shaylor]

I’d go a few steps further, though. Wiley suggests that “universities would still make money, though, because they have a marketable commodity: to get college credits and a diploma, you’d have to be a paying customer.” I’m not sure how things stand in the US, but here in the UK we have a saturation of graduates with qualifications that are either oversupplied or effectively irrelevant to obtaining a job (in parallel with a decline in the number of science and engineering graduates); as further education has become much more expensive (as a result of the government’s efforts to make it available to all, ironically) its final product has become devalued. What most employers want now is experience and demonstrable ability – two things that a diploma does not guarantee in any way.

So perhaps we’ll see a return to something like the old guild apprenticeship system, wherein people work for a company at the same time as they take an assortment of modular courses with direct relevance to the job in question, moving up the ranks as they gain – and demonstrate – the specialist knowledge and skills required, at the pace which best suits them. There’d be nothing to prevent someone learning beyond their discipline if they so chose, or spending a lifetime in pursuit of academic achievement.

In fact, the more I think about it, the more I’m put in mind of “Phaedrus’s university” as described in Zen and the Art of Motorcycle Maintenance (yes, I do have a hippie streak, as if you hadn’t guessed), the most important component of which is the way it decouples education from coercion, obligation and standardised achievement metrics. Pirsig’s ideas were considered pretty radical in their time, and largely dismissed as unworkable; in the light of the ever-growing ubiquity of the web and free content, maybe it’s time to take another look.

A fascinating article here at Physorg on how human beings remember and recall words. Researchers at Queensland University of Technology and the University of South Florida compare two ways of thinking about connections between similar words 1) Networks of similar words and 2) something analogous to spooky action at a distance:

…the researchers suggest that the probability of a word being activated in memory lies somewhere between Spreading Activation (in which words are individually recalled based on individually calculated conceptual distance) and Spooky Activation at a Distance (in which the cue word simultaneously activates the entire associative structure).

Most likely, Spreading Activation underestimates the strength of activation, while Spooky Activation at a Distance overestimates the strength of activation.

The researchers are using quantum physics as an preexisting abstract framework for their mathematical models for how human beings remember:

In the new model, associative word recall probability depends on how strongly connected the associated words are to each other.

For instance, “Earth” and “space” are entangled in the context of “planet,” but “Earth” and “gas giant” may not be entangled (though “Jupiter” and “gas giant” may be).

Words that are entangled with many other words have a greater probability of being recalled, while words that are entangled with few or no other words have a smaller recall probability.

At this stage this is theoretical, but the long-term consideration is for the development of AI and similar technologies:

As our information environment becomes more complex, we will need technology that can draw context-sensitive associations like the ones we would draw, but increasingly don’t as we lack the cognitive resources to do so.

Therefore, such the ‘meanings’ processed by such technology should be motivated from a socio-cognitive perspective.” This kind of research is an example of an emerging field called “quantum cognition,” the aim of which is to use quantum theory to develop radically new models of a variety of cognitive phenomena ranging from human memory to decision making.

Plenty of beef for the science-fictional burger bar.

[image from zeroinfluencer on flickr]