There is a huge difference between matching one name from a set of pre-recorded by the same speaker (normally not more than 10), as the case is in the phone, and recognizing more complex sentences, natural languages, from any speaker.
We developed several years ago a natural language MP3-player, where, using ID3-tags, you could say things like "play some romantic music" (matching on the genre), or "enqueue songs by Abba".
Lots of hidden issues, from voice user interface design, natural language parsing, making it multilingual etc, and having it run on embedded hardware. We got quite far though.
If you're interested, check out our company at www.voxi.com. Our VCs went bust, so there's not much going on at the moment, but we have some nice IP if anyone's interested:-)
From what I understand, Wi-Fi requires considerably more power than Bluetooth, and this might be a reason for phone manufacturers to go for BT instead.
I wonder if the Bluetooth connection can be used as an Internet link instead of GPRS...
I can't quite understand why Mozilla needs to use 40-100 MB. It would be interesting to see what that memory is actually used for. Images are normally fairly small, and can't take up *that* much. HTML source shouldn't account for much. Where does the memory go?
I wrote a mandelbrot program in postscript once. Tiny, but it took hours for the printer to print it, to the dismay of everyone behind me in the printer queue.
Actually, one thing I like about Gibson's style is how he uses invented terms - some you can guess from their morphology what they mean, others you can't.
So you form an idea of what each word seems to mean out of the context - like you do when you hear a novel word.
Finally most if not all of the new terms are explained a little while after they are first used, giving an interesting experience of how your deduced meaning matches the intended one.
One of the great unsolved problems of mathematics, the so called Riemann Hypothesis, may have a solution 144 years after Bernhard Riemann published his idea of a special equation, related to prime numbers.
If the proof does turn out to be correct, the two problem solvers may be eventually be able to collect a one million dollar reward.
In the year 1900, the world's leading mathematicians gathered for a conference in Paris. During the conference, David Hilbert, the leading mind of mathematics at the time, presented 23 problems which would affect mathematics for the ensuing century, and yet today. One of these problems was the Riemann hypothesis, and despite great effort it has remained unsolved. However, in November of 2002, Carlos Castro of Clark Atlanta University, Atlanta, USA, and Jorge Mahecha of the University of Antioquia, Medellin, Colombia published a proposed solution.
They won't be able to collect the one million dollar reward offered by the american Clay Mathematics Institute until one year after publication. This is to allow other mathematicians time to check the result, and verify its correctness.
- If this is true, then it is a sensational. This is a class of problems which requires much work before a possible solution can be confirmed. Thousands of mathematicians all over the world will cast themselves at this, and examine the evidence in minute detail, says Anders Karlqvist, mathematician, professor of informatics and head of the Polar Research Secretariat.
He explains that certain of Hilberts problems are problem areas rather than individual problems. Some have also been reformulated. If the Riemann Hypothesis is solved, then ten of the 23 problems are solved, seven are unsolved, five are partially solved, and one is no longer relevant.
According to Anders Karlqvist, Hilbert was a great man within his field, and a great period ended with him. He was the last man to have an overview of all the fields of mathematics.
Mathematics have developed very rapidly during the latest decades, thanks to an aid that the mathematicians of the early 20th century couldn't predict - the computer. The ever faster and larger computers of today can handle vast quantities of numbers and quickly make calculations that used to be impossible for a person even if he or she spent an entire lifetime.
With the help of computers, certain problems have been solved, such as the four color problem. It says that at most four different colors are neccessary to colour a map, so that areas with a common border don't have the same color. A computer program has systematically gone through all possible combinations.
Anders Karlqvist thinks that this involves a philosophical dilemma: should proofs in the form of computer programs be accepted? He believes that we stand before a culture shift within mathematics. During the coming decade mathematics will develop radically due to the increasingly efficient computers.
I agree that this is awesome technology with great impact.
Don't forget J2ME (Java 2 Micro Edition) which is becoming common in less expensive phones.
Although not nearly as powerful computationally, I think the combination of phones that can be programmed by anybody and GPRS packet data networking will be just as important as the more expensive 'communcator' phones, because it will reach the masses.
#Begin Nostalgia Hopefully J2ME programming on limited devices may bring back something of the art of efficient programming, which has been neglected on desktop machines with GHz, hundres of MBs of RAM, and Gigs of hard disks. Remember all that could be done on 1 MHz Apple IIs with 64 kB of RAM... #End Nostalgia
...but if you want to build a number of devices and actually use them (not just buying one kit and reprogramming it), then it gets prohibitively expensive.
FPGAs are really neat. There is so much that could be done on a hobby scale by programming hardware. It would be awesome to build many cheap network audio devices and spread them about.
However, the problem is just as stated, that there are no "motherboards" with FPGAs that are cheaper than regular computers. And all FPGA chip packages have such high pin density that it is not feasable for a hobbyist to solder a board, let alone many.
Alas, an awesome technology, but not reasonably cost-efficient for hobby projects.
Re:Anonymity: A problem with Gnutella
on
Gnutella2?
·
· Score: 1
Thanks for your comments!
You wrote:
The best solution is to have the Gnutella nodes work as routers. Each one passes the data off to the next.
Agreed. I think this is the only way it can be done.
Let us assume that A wants to dowload from B. Well there are a few nodes in between them (4 hops). Well, how is one going to no what the next hop is? Either A or B is going to need to know the complete path the the other.
With Internet routing, each node only knows what the next hop is, and doesn't have to know the entire path. I would think that we could port routing discovery algorithms in use on the Internet for routing between "Gnutella" nodes.
For that small improvement in anonymity, you are now downloading at 1/100th the speed you would get from a direct connection.
How did you arrive at that result (1/100th of the original speed)? It is not at all obvious to me what the speed impact will be.
If the routing is organized such that most of the routing nodes are on the backbone, then the speed impact for end nodes on slower connections would not have to be affected so much. It would more be a question of latency rather than speed being affected, which is not so important.
To reach this goal, a self-organizing routing scheme which can detect the speed (and maybe also hops) between nodes will be required.
However, the total amount of traffic on the Internet will increase, because all the data will be transmitted more than once. However, if the routing organization is good, this effect can hopefully be compensated for.
Additionally, the node working as the router directly connected to A or B could tell that they are the end nodes. So, on top of all that, you need encryption.
I think you could do this without the routers knowing if the next hop is the final one.
Each router only has local knowledge of where to go for the next hop. You could do something like this:
Nodes: A - B - C
A wants to fetch data from C. A can figure out that B is the next node on the path to C. A asks B to fetch a certain data from C (and gives this transaction a unique identifier).
B then asks C to get the named data, and send it back to B. C won't know if B is the final destination, or not. Similarly, B can't know if A is the final recipient for the data or not. B can use the stream identifier to know that the next hop for this packet is to A.
So as far as the transport protocol goes, I don't think you need encryption. However, maybe you would to prevent "bad guy" routing nodes to selectively drop certain data it is asked to route.
I would be happy to continue the discussion...
Anonymity: A problem with Gnutella
on
Gnutella2?
·
· Score: 1
A major problem for me with Gnutella is anonymity. I would like to be able to download and upload files without revealing my IP number.
I suppose it is debatable whether this is solvable without changing Gnutella's properties beyond recognition.
I know Freenet is an alternative, and I havn't tried the latest version, but I think Freenet would have problems implementing a search function like Gnutella's.
Some ideas of how to run something like Gnutella anonymously are:
* using something like the Anonymizer. However, I'd have to trust the anonymizer:-(
* Running over IPv6, getting an IPv6 address that is harder for the bad guys to track/convert to IPv4 than using a normal IPv4 address.
* Designing a protocol where data travels through several clients on its way from source to destination. Each client doesn't know if the next client is the final destination or not, the clients basically act as routers and only know the next hop to take on its way to its destination. This is similar but not equivalent to FreeNet (no encrypted resource identifiers, each node holds exactly the data which it serves (not encrypted)).
This would be less efficient than current Gnutella. It would require some reasonably efficient self-organizing routing scheme (this seems like the most difficult part of an implementation). But it could hopefully be more efficient than Freenet.
Personally, I would value anonymity enough to sacrifice a fair amount of performance.
The Nokia 6310i is a triband Bluetooth phone which has gotten good reviews.
As opposed to many of the Nokia phones, it is actually possible to buy this one.
Unfortunately, as I understand it, it is not possible to use Voice Dialling with any currently available Bluetooth headset and Nokia phones, which definitely sucks.
I'm using a Matrox G400TV. It handles both NTSC and PAL.
However, I think it's been discontinued unfortunately, with no TV tuner replacement card from Matrox.
Would that require a hardware upgrade of the SliMP3, or could it be an upgrade to existing devices?
I think such a feature would be great, and facilitate for instance playing synthesized speech from the computer system to the user.
I'm planning to buy a SliMP3 within a month (when a friend of mine is coming to Europe from the US; the shipping rates to Sweden were very high! About $65 I think).
I agree with the previous poster that we (Slashdot users) should be allowed to insist on the editors doing a good job.
If they don't (as in this case not correcting an obvious spelling error), I think the comments to the posting is a reasonable place to critisize also meta-issues, like the selection of articles to be posted, or spelling errors.
As far as I know, there is no other forum for discussing the work of the Slashdot editors.
Therefore, I think it is wrong for moderators to mod-down meta-comments as off-topic, as long as there is no other forum on Slashdot where it is on-topic!
Slashdot Mirror
The iPod Touch with a microphone as a VoIP WLAN phone would have been sweet.
Does anyone know what hardware the professional "photo printers" use?
What their costs are, and what volume is neccessary to motivate "professional" printers (whatever that is)?
What technology do they use?
The story states:
"The most common way in making hydrogen is electrolyzing pure water."
From what I understand, this is wrong. I've heard that most hydrogen is ironically produced as a byproduct of refining oil.
Wikipedia for instance http://en.wikipedia.org/wiki/Hydrogen says that:
"Commercial bulk hydrogen is usually produced by the steam reforming of natural gas."
There is a huge difference between matching one name from a set of pre-recorded by the same speaker (normally not more than 10), as the case is in the phone, and recognizing more complex sentences, natural languages, from any speaker.
:-)
We developed several years ago a natural language MP3-player, where, using ID3-tags, you could say things like "play some romantic music" (matching on the genre), or "enqueue songs by Abba".
Lots of hidden issues, from voice user interface design, natural language parsing, making it multilingual etc, and having it run on embedded hardware. We got quite far though.
If you're interested, check out our company at www.voxi.com. Our VCs went bust, so there's not much going on at the moment, but we have some nice IP if anyone's interested
Why not?
Actually, longer words and sentences are easier for computers to recognize because they differ more than short commands.
I've worked for a long time with this stuff.
From what I understand, Wi-Fi requires considerably more power than Bluetooth, and this might be a reason for phone manufacturers to go for BT instead.
I wonder if the Bluetooth connection can be used as an Internet link instead of GPRS...
I can't quite understand why Mozilla needs to use 40-100 MB. It would be interesting to see what that memory is actually used for. Images are normally fairly small, and can't take up *that* much. HTML source shouldn't account for much. Where does the memory go?
I had the impression that you could store your bookmarks etc in LDAP, is this not so?
Is that the CPU or the entire MoBo that consumes 2.5W?
It would be really cool if it were possible to build a somewhat big, but still portable device to 'prototype' things like MP3 players, etc.
I wounder what kind of power solutions you could do and how many hours of runtime you could get. Even if it is a 'backpack' computer.
Ok, so it would be similar to a laptop, but more customizable probably in terms of power configuration etc.
A space seems to have snuch into the URL between 2003 and -June. Remove it and the URL works. Here it is clickable: The Link
I wrote a mandelbrot program in postscript once. Tiny, but it took hours for the printer to print it, to the dismay of everyone behind me in the printer queue.
Actually, one thing I like about Gibson's style is how he uses invented terms - some you can guess from their morphology what they mean, others you can't.
So you form an idea of what each word seems to mean out of the context - like you do when you hear a novel word.
Finally most if not all of the new terms are explained a little while after they are first used, giving an interesting experience of how your deduced meaning matches the intended one.
Ok, I agree that's cool from a geek perspective, but I fail to see it as a "killer app" which will appeal to the public at large.
What value does broadband + VoIP deliver that the current phone network doesn't?
Here's a human translation:
Classical Math Problem May Be Solved
One of the great unsolved problems of mathematics, the so called Riemann Hypothesis, may have a solution 144 years after Bernhard Riemann published his idea of a special equation, related to prime numbers.
If the proof does turn out to be correct, the two problem solvers may be eventually be able to collect a one million dollar reward.
In the year 1900, the world's leading mathematicians gathered for a conference in Paris. During the conference, David Hilbert, the leading mind of mathematics at the time, presented 23 problems which would affect mathematics for the ensuing century, and yet today. One of these problems was the Riemann hypothesis, and despite great effort it has remained unsolved. However, in November of 2002, Carlos Castro of Clark Atlanta University, Atlanta, USA, and Jorge Mahecha of the University of Antioquia, Medellin, Colombia published a proposed solution.
They won't be able to collect the one million dollar reward offered by the american Clay Mathematics Institute until one year after publication. This is to allow other mathematicians time to check the result, and verify its correctness.
- If this is true, then it is a sensational. This is a class of problems which requires much work before a possible solution can be confirmed. Thousands of mathematicians all over the world will cast themselves at this, and examine the evidence in minute detail, says Anders Karlqvist, mathematician, professor of informatics and head of the Polar Research Secretariat.
He explains that certain of Hilberts problems are problem areas rather than individual problems. Some have also been reformulated. If the Riemann Hypothesis is solved, then ten of the 23 problems are solved, seven are unsolved, five are partially solved, and one is no longer relevant.
According to Anders Karlqvist, Hilbert was a great man within his field, and a great period ended with him. He was the last man to have an overview of all the fields of mathematics.
Mathematics have developed very rapidly during the latest decades, thanks to an aid that the mathematicians of the early 20th century couldn't predict - the computer. The ever faster and larger computers of today can handle vast quantities of numbers and quickly make calculations that used to be impossible for a person even if he or she spent an entire lifetime.
With the help of computers, certain problems have been solved, such as the four color problem. It says that at most four different colors are neccessary to colour a map, so that areas with a common border don't have the same color. A computer program has systematically gone through all possible combinations.
Anders Karlqvist thinks that this involves a philosophical dilemma: should proofs in the form of computer programs be accepted? He believes that we stand before a culture shift within mathematics. During the coming decade mathematics will develop radically due to the increasingly efficient computers.
Bengt Jonsson
I agree that this is awesome technology with great impact.
Don't forget J2ME (Java 2 Micro Edition) which is becoming common in less expensive phones.
Although not nearly as powerful computationally, I think the combination of phones that can be programmed by anybody and GPRS packet data networking will be just as important as the more expensive 'communcator' phones, because it will reach the masses.
#Begin Nostalgia
Hopefully J2ME programming on limited devices may bring back something of the art of efficient programming, which has been neglected on desktop machines with GHz, hundres of MBs of RAM, and Gigs of hard disks. Remember all that could be done on 1 MHz Apple IIs with 64 kB of RAM...
#End Nostalgia
...but if you want to build a number of devices and actually use them (not just buying one kit and reprogramming it), then it gets prohibitively expensive.
FPGAs are really neat. There is so much that could be done on a hobby scale by programming hardware. It would be awesome to build many cheap network audio devices and spread them about.
However, the problem is just as stated, that there are no "motherboards" with FPGAs that are cheaper than regular computers. And all FPGA chip packages have such high pin density that it is not feasable for a hobbyist to solder a board, let alone many.
Alas, an awesome technology, but not reasonably cost-efficient for hobby projects.
You wrote:
Agreed. I think this is the only way it can be done.With Internet routing, each node only knows what the next hop is, and doesn't have to know the entire path. I would think that we could port routing discovery algorithms in use on the Internet for routing between "Gnutella" nodes.
How did you arrive at that result (1/100th of the original speed)? It is not at all obvious to me what the speed impact will be.
If the routing is organized such that most of the routing nodes are on the backbone, then the speed impact for end nodes on slower connections would not have to be affected so much. It would more be a question of latency rather than speed being affected, which is not so important.
To reach this goal, a self-organizing routing scheme which can detect the speed (and maybe also hops) between nodes will be required.
However, the total amount of traffic on the Internet will increase, because all the data will be transmitted more than once. However, if the routing organization is good, this effect can hopefully be compensated for.
I think you could do this without the routers knowing if the next hop is the final one.Each router only has local knowledge of where to go for the next hop. You could do something like this:
Nodes: A - B - C
A wants to fetch data from C. A can figure out that B is the next node on the path to C. A asks B to fetch a certain data from C (and gives this transaction a unique identifier).
B then asks C to get the named data, and send it back to B. C won't know if B is the final destination, or not. Similarly, B can't know if A is the final recipient for the data or not. B can use the stream identifier to know that the next hop for this packet is to A.
So as far as the transport protocol goes, I don't think you need encryption. However, maybe you would to prevent "bad guy" routing nodes to selectively drop certain data it is asked to route.
I would be happy to continue the discussion...
A major problem for me with Gnutella is anonymity. I would like to be able to download and upload files without revealing my IP number.
:-(
I suppose it is debatable whether this is solvable without changing Gnutella's properties beyond recognition.
I know Freenet is an alternative, and I havn't tried the latest version, but I think Freenet would have problems implementing a search function like Gnutella's.
Some ideas of how to run something like Gnutella anonymously are:
* using something like the Anonymizer. However, I'd have to trust the anonymizer
* Running over IPv6, getting an IPv6 address that is harder for the bad guys to track/convert to IPv4 than using a normal IPv4 address.
* Designing a protocol where data travels through several clients on its way from source to destination. Each client doesn't know if the next client is the final destination or not, the clients basically act as routers and only know the next hop to take on its way to its destination. This is similar but not equivalent to FreeNet (no encrypted resource identifiers, each node holds exactly the data which it serves (not encrypted)).
This would be less efficient than current Gnutella. It would require some reasonably efficient self-organizing routing scheme (this seems like the most difficult part of an implementation). But it could hopefully be more efficient than Freenet.
Personally, I would value anonymity enough to sacrifice a fair amount of performance.
Comments?
As opposed to many of the Nokia phones, it is actually possible to buy this one.
Unfortunately, as I understand it, it is not possible to use Voice Dialling with any currently available Bluetooth headset and Nokia phones, which definitely sucks.
I'm using a Matrox G400TV. It handles both NTSC and PAL.
However, I think it's been discontinued unfortunately, with no TV tuner replacement card from Matrox.
Would that require a hardware upgrade of the SliMP3, or could it be an upgrade to existing devices?
I think such a feature would be great, and facilitate for instance playing synthesized speech from the computer system to the user.
I'm planning to buy a SliMP3 within a month (when a friend of mine is coming to Europe from the US; the shipping rates to Sweden were very high! About $65 I think).
What does paying someone money have to do with him/her doing a good job?
Especially in the context of the open source / free software movement.
I agree with the previous poster that we (Slashdot users) should be allowed to insist on the editors doing a good job.
If they don't (as in this case not correcting an obvious spelling error), I think the comments to the posting is a reasonable place to critisize also meta-issues, like the selection of articles to be posted, or spelling errors.
As far as I know, there is no other forum for discussing the work of the Slashdot editors.
Therefore, I think it is wrong for moderators to mod-down meta-comments as off-topic, as long as there is no other forum on Slashdot where it is on-topic!
(Puts on the Asbestos suit)