Archive for the ‘Buzzwords’ Category

How chaos could improve speech recognition

Wednesday, December 24th, 2008

a-sound

If you’ve ever used speech recognition software, you’ll know how often it fails to work well. Recognition rates are nowhere near what is needed for anything but the simplest applications.

So a new approach for analysing speech by Yuri Andreyev and Maxim Koroteev at the Institute of Radioengineering and Electronics of the Russian Academy of Sciences in Moscow is welcome. Their approach is to treat the production of speech as a chaotic phenomenon.

That’s a significant difference compared with previous approaches which predict the next point in a speech signal by extrapolating from previous points in a linear fashion.

That works because the organs that produce speech–the vocal cords–change over a much longer time period than the sound they produce. So they can be considered essentially stationary for this type of analysis.

Of course, one of the characteristics of chaos is that very small changes in starting conditions can produce large changes in output. And if that’s happening, what kind of chaos are we talking about?

Andreyev and Koroteev answer this question by measuring the frequency and amplitude of the sound a person makes when saying various vowels and consonants. They then use this data to reconstruct the multidimensional phase space in which the chaotic signal is produced.

The results are interesting because specific vowels appear to be linked to unique structures in the phase space. Andreyev and Koroteev call these structures phase portraits. The picture above is a phase portrait of the vowel sound ‘a’.

It’s a little harder to identify the shapes associated with consonants and the researchers haven’t yet tried with other sounds such as dipthongs.

It’s a long step from here to speech recognition but in principle, it could be done by looking for the phase portraits of specific phonemes and using them to spell out words.

The question, of course, is whether this would be easier or harder than current approaches.

Ref: arxiv.org/abs/0812.4172: On Chaotic Nature of Speech Signals

How chaos could improve speech recognition

Wednesday, December 24th, 2008

a-sound

If you’ve ever used speech recognition software, you’ll know how often it fails to work well. Recognition rates are nowhere near what is needed for anything but the simplest applications.

So a new approach for analysing speech by Yuri Andreyev and Maxim Koroteev at the Institute of Radioengineering and Electronics of the Russian Academy of Sciences in Moscow is welcome. Their approach is to treat the production of speech as a chaotic phenomenon.

That’s a significant difference compared with previous approaches which predict the next point in a speech signal by extrapolating from previous points in a linear fashion.

That works because the organs that produce speech–the vocal cords–change over a much longer time period than the sound they produce. So they can be considered essentially stationary for this type of analysis.

Of course, one of the characteristics of chaos is that very small changes in starting conditions can produce large changes in output. And if that’s happening, what kind of chaos are we talking about?

Andreyev and Koroteev answer this question by measuring the frequency and amplitude of the sound a person makes when saying various vowels and consonants. They then use this data to reconstruct the multidimensional phase space in which the chaotic signal is produced.

The results are interesting because specific vowels appear to be linked to unique structures in the phase space. Andreyev and Koroteev call these structures phase portraits. The picture above is a phase portrait of the vowel sound ‘a’.

It’s a little harder to identify the shapes associated with consonants and the researchers haven’t yet tried with other sounds such as dipthongs.

It’s a long step from here to speech recognition but in principle, it could be done by looking for the phase portraits of specific phonemes and using them to spell out words.

The question, of course, is whether this would be easier or harder than current approaches.

Ref: arxiv.org/abs/0812.4172: On Chaotic Nature of Speech Signals

A clue in the puzzle of perfect synchronization in the brain

Thursday, November 27th, 2008

zero-lag-synchronisation

“Two identical chaotic systems starting from almost identical initial states, end in completely uncorrelated trajectories. On the other hand, chaotic systems which are mutually coupled by some of their internal variables often synchronize to a collective dynamical behavior,” write Meital Zigzag at Bar-Ilan University in Israel and colleagues o the arXiv today.

And perhaps the most fascinating of these synchronized systems are those that show zero lag; that are perfectly synched. For example, in widely separated regions of the brain, zero lag synchronization of neural activity seems to be an important feature of the way we think.

This type of synchronization also turns out to be an important feature of chaotic communication. This is the process by which which information can be hidden in the evolution of a chaotic attractor and retrieved by substracting the same chaotic background to reveal the original message.

Obviously, this only works when the transmitter and receiver have are coupled so that they evolve in exactly the same way. For a long time physicists have wondered whether this effect can be used to send data securely and earlier this year, they proved that the security can only be guaranteed if the synchronisation has zero lag.

But how does zero lag occur and under what range of conditions?

Zero lag seems to occur when the delays in the mutual coupling and self feedback between two systems act to keep them in step. In effect, both systems lag but by exactly the same amount.

Until recently, this was thought to occur only for a very small subset of parameters in which the delays are identical or have a certain ratio. But these limits are so exact and constricting that it’s hard to imagine a wet system such as the brain ever achieving them.

Now Zigzag and friends have shown that it is possible to get around these strict limits by having more than one type of feedback between the systems. When that happens, it’s possible to have zero lag synchronisation over a much wider set of parameters.

That’s going to have important implications for our understanding of synchronisation in the brain and for the development of secure chaotic communication. Betcha!

Ref: arxiv.org/abs/0811.4066: Emergence of Zero-Lag Synchronization in Generic Mutually Coupled Chaotic Systems

How to build a quantum internet

Monday, June 30th, 2008

Quantum internet

You could be forgiven for thinking that a quantum version of the internet is a couple of-afternoons-in-the-lab away from being plumbed into your living room. In reality, there are significant engineering challenges to overcome, says Jeff Kimble from the California Institute of Technology in Pasadena and one of the leading thinkers on the links between the quantum and information sciences.

If you want to know about some of the bigger hurdles that physicists face in wanting to build a quantum internet, you could do worse than look at his account of this field on the arXiv today.

Some of the challenges are particularly daunting such as the unambiguous creation and verification of entanglement.

But, ever the optimist,  he concludes:

“I have every confidence that extending entanglement across quantum networks will create wonderful scientific opportunities for the exploration of physical systems that have not heretofore existed in the natural world”.

Ref: arxiv.org/abs/0806.4195: The Quantum Internet

Solar system filled with dark matter, say astronomers

Thursday, June 26th, 2008

Solar dark matter

As the evidence for dark matter builds, astronomers have begun modelling how it ought to be distributed around the cosmos.  They’ve shown how it must be distributed on the largest scale to make clusters of galaxies form in the way we see, various other simulations show that it forms a kind of halo around galaxies such as the Milky Way.

But what of smaller scales? Today, Ethan Siegel and his student Xiaoying Xu at the University of Arizona produce the first model showing how much dark matter there is in the Solar System.

They say that throughout its 4.5 billion year history, the Sun will have been sweeping up the dark stuff as it has moved through the Milky Way. Siegel estimates that in this time it will have gathered some 8 x10^19 kilograms of dark matter, or about 300 times the background levels in the Milky Way.

That’s a lot of mass and Siegel points out that that much dark matter should have profound implications for the various teams searching for it. It means there ought to be more dark matter than anybody expected although this will have a smaller velocity relative to Earth because it should be moving through the galaxy with the Sun. Keep ’em peeled, guys.

It’s also easy to think that this much dark matter  might have a bearing on the Pioneer anomaly, the unexplained acceleration towards the Sun of our most distant space probes. But no, says Siegel.

This amount of dark matter is much several orders of magnitude smaller than  than the mass of Pluto or any of the larger moons in the solar system. So there’s not nearly enough to explain the observed accelerations, he concludes.

So that’s that then.

Ref: arxiv.org/abs/0806.3767: Dark Matter in the Solar System

The birth of piezo-spintronics

Wednesday, December 5th, 2007

Spin polarizer

Everything we do with electrons depends on their charge, for example the whole of electronics is based around an electron’s negative charge. But electrons have another property, their spin, that could also be used to manipulate them. So a lot of engineers are trying to think up ways we can fiddle with an electron’s spin to do interesting things. The resultant field is called spintronics and it’s exciting cos it opens the possibility of an entirely new form of computing based on spin rather than charge.

But it ain’t just computing that depend son an electron’s charge. Piezoelectric materials also depend on the way charge is distributed within their crystalline structure. When squeezed, this symmetry changes in way that generates a voltage. Conversely, when a voltage is applied, the crystals themselves change shape.

Now Alexy “Koala” Kovalev at Texas A&M University at College and some egghead pals are proposing that electron spin can do a similar thing. Pass a spin-polarized current (one in which all the electron spins point in the same direction) through the right kinda crystal and it will bend or vibrate. That ain’t so useful cos we can make things vibrate pretty good now.

But the converse effect could be very useful. Send a current through an oscillatin’ wire and it’ll spin-polarize it. Koala is calling it the piezo-spintronic effect and it’s handy cos there ain’t that many ways to generate spin-polarized currents right now and none that look as simple as this.

One problem: Koala Kovalev and pals have done some neat thinkin ‘n’ theorizin’ but ain’t built nothin’ yet. The chances are that this kinda piezo-spintronic material ain’t gonna be easy to come by. Time for some metamaterial brainstorming.

Ref: arxiv.org/abs/0711.4430: Nanomechanical Spin-polarizer

The birth and death of spookytechnology

Monday, October 22nd, 2007

Quantum mechanics is seriously sexy . That’s what keeps the public coming back for more with the drool hangin from their spittle-flecked lips (repeat this often enough and it might actually come true).

Without a sexy lingo to bamboozle and baffle, physics eggs ain’t got nothing to keep people’s attention ‘cept their rugged good looks and scintillatin’ personalities.

So labs hire hampers of postdocs whose sole task is to invent new words that’ll keep the public a-gazin’ and a-gawpin at every shiny new pebble that is thrown their way.

Yer probably didn’t realise it but all the latest buzzwords were invented in this way: atomtronics, quamputing, brane theory and the ballsheet model of particle physics (OK, ah made that one up).

Normally, these postdocs are thrown a stale crust to keep em happy but otherwise kept under strict lock and key. But this week, one of em has escaped and posted a buzzword on the arXiv without permission.

Charles “Manhat” Tahan is normally locked in the bowels of the Cavendish lab in Cambridge. Let loose, he is brayin’ that physicists are a-stuttering and a-stumblin for want of a word to describe technology that depends on quantum effects that have no classical counterpart. We’re talkin things like entanglement, coherence and superposition–the so-called spooky properties of quantum physics.

Manhat Tahan says the word we need is “spookytechnology”.

Whadaya’ll think? Don’t send no shivers down my spine, I gotta tell yer. Surely ya’ll can come up with something better for this kinda strange stuff. How about quantech or ungodly gear or maniacal machinery or EPR matter or subsubstance or whatever.

And then maybe, just maybe, we can persuade Manhat Tahan to get back into the dungeon before he does some real damage.

Ref: arxiv.org/abs/0710.2537: Spookytechnology and Society

Magnetic cloaking

Wednesday, October 17th, 2007

The world has gone crazy over metamaterials cos they can be used to build invisibility cloaks, as ya’ll saw just the other week. There’s usually some drawback the media coverage never tells ya which means that we ain’t gonna see no Harry Potter-type invisibility cloaks any time soon. But that hasn’t stopped the living God of metamaterials, John “Now-You-See-Him” Pendry at Imperial College in London UK, from dreamin up ever more ingenious designs.

This week, he and a few pals have gone and built a metamaterial that works for light with zero frequency. That’s egghead-talk for a magnetic field. The material is a thin layer of lead squares pasted onto a cover slip and cooled to superconducting temperatures. This actually prevents a magnetic field from passing through it.

Why would anybody need a metamaterial that screens magnetic fields? There are all kindsa researchers playin around with tiny magnetic fields or with experiments in which they don’t want no magnetism at all. Their work is swamped by larger fields nearby. So a magnetic field cloak might be a useful device to have around.

And unlike other cloaking ideas, the technology is up and running now.

Ref: arxiv.org/abs/0710.1227: A DC Magnetic Metamaterial

Crowdquakes–the killers that cause stampedes

Tuesday, September 4th, 2007

The squeeze ‘n’ shove of Mecca pilgrimages are a-mighty frightenin. Thousands of people have been died in em. Now video analysis shows that a remarkable new phenomenon called crowdquakes are behind these stampedes. The analysis  also suggests how stampedes might be prevented.

Dizzy Dirk Helbing and his mates at the Swiss Federal Institute of Technology in Zurich have been a-studyin videos of crowd panic and stampedes at the Jamarat Bridge near Mecca, where 350 pilgrims were trampled to death in 2005 (250 also died in 2004).

Conventional models of crowd behaviour assume that the velocity of the flow is smooth and that it drops to zero as the density of people rises–in other words the crowd drifts to a halt.

That ain’t what happens at all though. Dizzy Dirk reckons that at densities of more than 7 people per square meter, an individual is no longer able to control his movement and the crowd begins to move as a whole.

The density within crowds varies from place to place. So where the density is above the threshold, whatcha get is regions that take on a life of their own, a-heavin and a-shovin and a- sendin pressure waves through less dense regions of the crowd.

Now here’s the interestin bit. Dizzy Dirk says that under these circumstances, chains of people can become so tightly compressed that they become momentarily locked, like sand jamming as it passes through an hour glass. When these force chains break, they release sudden, uncontrollable amounts of energy, like earthquakes.

It is these crowdquakes that knock people off their feet causing them to be trampled. Dizzy Dirk says he seen it happen in the Jamarat videos.

The solution? Prevent crowd densities exceeding 7 bodies per square meter. In the videos of the 2006 event in which a crowdquake killed more than 350 people, Dirk says the warning signs were clearly visible 11 minutes before it occurred. That should give the authorities some warning.

Ref: arxiv.org/abs/0708.3339: Crowd turbulence: the physics of crowd disasters