Slashdot Mirror

What does speech recognition have to do with the Turing test?

Proper speech recognition has been proven to be AI-complete, similar to "The Vision Problem" (building a system that can see as well as a human), and many others. Perhaps not proven as rigorously as mathematical theorems but all data is pointing this way.
Therefore, correctly solving the speech recognition problem is equivalent to solving the Turing test. So if anybody predicts good speech recognition in some near future, it is usually a sign of uninformedness and that person probably shouldn't be taken seriously.

Be quite concerned about Ken Thompsons idea. now shouldnt they?

The very first "micro" (as we called them back in the day) I owned was a Sinclair ZX80 which was marketed as being up to the job of managing a nuclear power plant! Maybe that could be proved true, in a way, by running it off one of these things!

Since he teaches computer science at the university level, he had better know the difference between an operating system and user interface. Gelernter says:

An operating system connects the user (and the user's software) to the ensemble of machines we call a computer.

So far, so good.

Unfortunately, he goes on to immediately add:

But nowadays users no longer want to be connected to computers. They want to be connected to information

This blurs the distinction between OS and UI, and implies that the main job of the operating system is to somehow mystically "connect" users to the computer. It makes the human user sound like a peripheral. It is a simplification that borders on demagoguery. Yes, people want information. But something or someone has to handle the low-level tasks of co-ordinating all those 1s and 0s. Those users who don't want to be "connected" to computers certainly would not want to spend their time on this. It would be better to hand this job off to software. Gee, that sounds like an operating system.

The good news, for prospective purchasers, is that the range on this thing is pretty decent. Not sure what the bluetooth spec quotes for range, but it works perfectly in the 50 - 60ft length of the theatres. The bad news is that this means lecturers feel far more confident about walking around and asking questions of the insomniacs catching up on some much needed shuteye at the back.

Has anyone had any experience with two of these devices in the same bluetooth hotspot? Not that I've got anything planned, at all...

OK, the too many movies line is getting old. Let me flesh this out a bit.

I graduated from the United States Naval Academy with a Degree in Engineering and have been invited back to teach Computer Science by that department's chair. I am currently living in Japan where I study Distributed Operating System design and Artificial Intelligence research in my spare time. While I may speak enough Japanese to pick up a girl at a bar, I do not understand enough to watch movies; hence why I don't have cable.

The FACTs of the discussion are these:

#1 Research on the human brain and how it learns is still considered to be in its infancy. Current experiements in the subject are almost barbaric in their simplicity; such as performing CT scans on a subjects brain while they are exposed to extremem emotional states, just to see which location in the brain becomes more active.

#2 Expert systems already exist today that are capable of creating, or rewriting the rules they operate off of. Exactly how much memory do they need to have and how much information do they need before they begin writing rules that don't pertain to their original design? What point does the information processing load become more than the current hardware can handle?

#3 Neural networks of some pretty fair complexity have been played with, but we still haven't quite reached the stage where the computer will mimic human learning and no one can say how close we are. Again, how do we know we don't already have the right algorithm, and just not the right input information.

One might infer from my sig that I read too many books, but Dostoevsky had the fall of the Russian government, the rise of communism and the abolishment of the church in Russia nailed decades before it actually happened. Bullishly pushing forward with a misunderstood technology or concept has screwed us over many times. I don't think it's out of line to accept that the artistic half of our species occassionally has very clear insight into our future.

PowerOpen would be a suitable ABI for any UNIX app that's POSIX compliant. Perhaps it could even be extended to work with cocoa apps?

I hate to break it to you, but OS X is so slow on the G3 that you might as well not bother.

Not at all. Time to drop your FUD. I just recently aquired a used 400 Mhz G3 PowerBook with 256MB RAM, and things are running on it quite well. iTunes, iCal, Mozilla, AOL, etc all at once. I've even been doing a lot of remote work with OroborOSX running apps from my Linux box remotely, including Mozilla and full-bore developer stuff.

(I'm quite interested in looking into Rendezvous also, given that it's a Zeroconf implementation.)

EBCDIC has won yet another chess game against the Grim Reaper.

told me unequivocally that AOL is the Internet.

Well strictly, AOL is an internet, which is also part of The Internet. It still amazes me how many journalists get the capitalization wrong! :-)

told me unequivocally that AOL is the Internet.

Well strictly, AOL is an internet, which is also part of The Internet. It still amazes me how many journalists get the capitalization wrong! :-)

Of course, making it easy to switch providers doesn't profit the providers- especially when their business models (in the US) rely on selling $40/month contracts, rather than affordable prepaid service, as is available in Europe. (Note that most prepaid options here in the US are expensive as hell.)

CDMA is the way to go for the RF end of things, at least until the Next Great Innovation is made. Qualcomm really did put a fair bit of effort into commercializing the tech, so I can't fault them too much; by choosing to downplay their own products and move into pushing patents, they're an example of a relatively 'non-evil' intellectual property company... sort of a Rambus who developed something the world might actually want.

SIM cards- or equivalent- are the way to go for the user interface ('consumer interface?'), but as the author points out, it's hard to say if the US could ever legislate the companies into providing them without killing the climate that let CDMA develop in the first place.

Personally, I like CDMA as a technology, and I like consumer choice. I suppose if this proves anything, it's that patents are a bit double-edged, these days; by the time Qualcomm's expire, 9600/14.4k transfers will be so far obsolete that the world won't benefit from a reimplementation (the available bandwidth will be shifted elsewhere)... if they'd somehow come to success in a patent-free world, they would've had to remain an equipment company, and continue competing with other manufacturers... but then TDMA chipsets would've had the same advantage *and* a head start, relegating the superior (and spectrum-conserving) protocol to an also-ran standard, sort of like what happened in the selection process for 100baseT- see the following link:

What happened in the selection process for 100baseT.

Of course, sticking with CSMA/CD probably did keep Ethernet chipsets from becoming as messy as HomePNA ones. *cough*petitionBroadcomtoopenHomePNA2.0specs*cough *

Hey, if you know a new solution to deadlocks and race conditions so that it's trivially easy to solve all of them in realtime, then go talk to a processor vendor of your choice - you won't ever have to invent anything again.

Until that happens it's simply not possible for anything but the most trivial of tasks (which is already done by compilers and processors with multiple execution units).

Hey, if you know a new solution to deadlocks and race conditions so that it's trivially easy to solve all of them in realtime, then go talk to a processor vendor of your choice - you won't ever have to invent anything again.

Until that happens it's simply not possible for anything but the most trivial of tasks (which is already done by compilers and processors with multiple execution units).

Hey Grammar nazis, go to oxford.ac.ic.uk

LoL. Where the fuck is that?

If you're talking about Oxford University then refer to it by name, if you're talking about its domain name, then GET IT RIGHT.

ic.ac.uk!=oxford.ac.uk

I have the perfect solution for you! It's something new that allows you to use your calendar while sitting in front of any computer; you can view it, add appointments, to-do items, or free-form notes quickly and easily, with total privacy and security, using an amazingly intuitive user interface!

Check it out!

(HHOS)

Finally I realized memetic systems defining reinsurance networks in terms of kin-selection were the most natural way to make stable technological civilizations because other memetic systems (those that deny the importance of kin-selection) merely evolve hypocrisy at an unconsciuos genetic level rendering rational thought, communication and action nonviable.

You don't have to. You simply download the XFree86 package for MacOS X and click on it to install. If you like your X to have all that Aqua gooeyness, download the excellent OrborOSX package & perform a single-click install.

To upgrade to Jaguar, download the new XTerm package update (1MB - whoopee). One click and you're done again!

No rpm -u, no make install, no gcc, no fink, no X11 source, no dev env. Now, how difficult is this??

An XPilot map about the Mad Cow Disease scandal.

http://www.doc.ic.ac.uk/~np2/xpilot/bse.html

I totally agree with you. A source package keeps everything needed to build a program in one file.

There's another easy way for unpacking a rpm (or most other archive formats into it's own directory with unarc foobar.rpm
Get unarc, a simple perl script from http://www.doc.ic.ac.uk/~epa98/work/apps/unarc/

Well, actually, if the connection would not be "proxy based" - it could make a lot of sense - to have a always-on (even low bandwidth) connection in your pocket which does not suck your batteries in a second. But this "proxy based concept" seems to be the fun-spoiler and aimed to make your life more difficult when you actually would like to do something fun with it. Like making it more difficult to get an IP masquared for that.

Mozilla

The open source web browser , designed for standards-compliance, performance, and portability, whose development is coordinated by mozilla.org.

The Mozilla project started in March 1998 when Netscape Communications Corporation released the source code of Netscape Communicator . The now abandoned version based on that code is referred to as "Mozilla Classic". Since then, much has been rewritten, including the layout engine, the networking library, and the front-end.

mozilla.org was set up by Netscape in January 1998 to coordinate development and to provide a point of contact for interested people.

Although a lot of Mozilla code is under the original Netscape Public License, some parts of the code are under the Mozilla Public License or dual MPL/GPL.

"Mozilla" was the original project code name for Netscape Navigator and, according to some of the documentation, the correct pronunciation of "Netscape".

Home (http://www.mozilla.org/) .

[Derived from "Mosaic killer/Godzilla"?]

Notice the last line?

This definition mentions such concepts as tables, relations between those columns, joins, etc. RDBMS is not an elitist term, as far as I can see -- it's a description of a class of databases. Other kinds of database exist: bdb (file-based hash), ZODB (an object store), hierarchical databases, and others. MySQL isn't one of those -- it is a relational database. It has tables, it has joins (even if integrity isn't ensured), it has a query language (that is not imperative)... it isn't a terribly featureful example, it does not pass the ACID test, but it is a relational database. A lot of the features given aren't even part of the original concept of an RDBMS (stored procedures in particular).

If MySQL is not relational, then what is it?

might want to be aware of this little feature at their disposal. Here on SlashNET, we frown at such things. I, as the ircd maintainer, refuse to include such things in our ircd and have refused patches such as this in the past. Evil stuff.

API is not Application-Programmer Interface: it's Application Programming (or Program) Interface; that is, the interface you use when you're programming an application.

Dijkstra was one of the pioneers in computer science. In most CS courses you can learn about Dijkstra work on semaphores and concurrency.

The Dining Philosophers problem is a famous synchronization problem proposed by Dijkstra concerning resource allocation between processes.

Here you can see some of his papers.

Well I got my degree in CS from one of the top academic CS institutions in europe (the world?) Imperial College and it's a MEng (i.e. an engineering degree). On the other hand, I remember reading a quote from someone famous (I forget who) saying CS was more like a cross between Maths (logic), Art (creativity & expression) and Biology (evoloution, interactions of complex systems). So I have no idea what it is :)

It's a shame that /. seems to think "Go To Considered Harmful" is Dijkstra's signature achievement. He was profoundly influential in developing the theory of operating systems. He was one of the first proponents of layered design. He also did pioneering work in mutual exclusion (IIRC, he invented semaphores) and deadlock. In short, he is responsible for a lot of the fundamental concepts that we use to build complex systems today.

MC Polynomial.

And he sang a song..."P can't Touch This". Before the drum break of this song, Poly sang:

It's a prime, because you know
P can't touch this
P can't touch this
Stop...Polynomial Time...

Thus giving rise to a branch of mathematical order functions denoting the complexity of a problem...

Either that or it's defined pretty well here.

Please read an entire post before replying.

Finally, what do I use for multithreaded application debugging? I don't have a tool for once I've cut the code but if you define your system using Finite State Processes then you can use the LTS Analyser to check for deadlocks etc and you can step through a concurrent system generating whatever event you like. It is worth a look, although matching your code to your model is still tricky. And of course you need to learn FSP, but the website above teaches you.

whoa... that guys hair is just wrong... or scary.. or both

This is what happens if you dont!

Or you culd catch them and make them pay for their evil deeds

Why do people always feel the need to not translate?

Because they post when they are in serious hack mode and make the assumption that others will understand without explinations.

Mea Culpa ;-)

Edinburgh doesn't teach computer science any more. It teaches Word Processing with How To Throw Together A Few Lines Of Code. That's brought the number of UK universities which actually teach proper computer science down from four to a rather paltry three. It's a great shame, really; without a pretty decent knowledge of computing as a science, any kind of software engineering is pretty tricky.

This is a good place to mention a project I am working on. I wasn't going to announce it quite yet because it's unfinished, but this JPEG news could change things.

I have modified zlib to support a simple form of lossy compression for images. The Lempel-Ziv compression stage works by looking for repeated strings in the data, and when a string is found just output a reference back to the previous occurrence. But what if this string matching allowed some deviation? More strings would be matched and the output file would be smaller, but there would be some loss of quality.

My implementation tries to do a kind of 'dithering string match' where it finds the longest possibel match where the total pixel difference (sum of squared differences of pixel values) is less than some given threshold. But the error in one pixel is taken into account when matching the next, so that one pattern of dithered pixels has a reasonable chance of matching another.

At present I just have the modified zlib and I am using it to compress 8-bit and 24-bit PNM images lossily. That is, with some reduction in quality I can compress a PNM image down to say 20 kilobytes, while gzip would compress it to 50 kilobytes. By adjusting the threshold you can trade off quality against file size (with threshold=0 you have plain gzip compression). The space saving and quality loss depend on the type of image being compressed. But the results so far are encouraging; with a bit more ingenuity in the matching algorithm it should be possible to do better.

To recap, so far I just have a modified zlib compressing 8-bit and 24-bit PNM images. I am trying to extend libpng to use this lossy zlib, but the trouble is that PNG headers should not be compressed lossily, just the pixel data. Also I need to deal with images that have more than one pixel per byte. (Although the results are still fairly good if you take a paletted image and convert it to 8-bit greyscale before lossy compression.)

I have quickly put together a web page Lossy PNG to demonstrate what has been done so far. I'm going away on holiday soon so I won't be able to make a working release for a while, but perhaps this proof of concept will encourage others to work on the idea.

First of all I want to point out that Apple tries to bring powerful technology to the masses by reducing unnecessary complexity. Something I don't look down on. In fact, I get annoyed when something is more complex than it has to be. That focus means, however, that they will hype things like iTunes, not the included Perl 5.6. On the other hand, they seem to be very hard at work at creating the best possible environment for developers, something which indeed was lacking in the past. But I digress and offer my assistance:

To get to the page you seek, you go to Apple.com and click on 'developer' (top right). The page that comes up contains links to documentation and sample code for when you decide to try out Cocoa, a login button for the developer programs (there is a free one that allows you to download the developer tools and other stuff) and an intro-page for programmers. You'll be very interested in that page since it contains links to various documents (the 'essential reading') on the technical underpinnings of OS X. You'll want to read up on that to understand the way things work. Next stop might be the page specific to Unix. You'll see that a few major Unix-apps are being named, but alas, no mention of python or tcl/tk. Fortunately, the page does contain a link to the Fink package manager (based on apt-get). You can find many more packages there, including a (optionally) rootless version of XFree. It allows you to use your familiar X apps next to OS X apps (be sure to use Orobor to have your X apps use Aqua widgets).

I don't have a list of all the Unix software that is standard on OS X or that you can download, but I don't think that you'll find any of the more popular tools missing. Given the number of Unix developers and users switching to OS X, I expect OS X to become a first class citizen in the Unix world quite soon. As of yet, you might still have to change paths in makefiles to get the less common software to configure, make, make install. I don't think an experienced Unix-guy like you will have a lot of problems with that however. In return for these minor issues, you will (hopefully) have a great experience using OS X.

Yup.

Look at http://foldoc.doc.ic.ac.uk/foldoc/foldoc.cgi?prefi x. And memorize it this time ;)

Javascript was previously known as Livescript.

The keyboard is odd Pet peeve: Home, End, Page Up and Page Down don't do what you expect - they are nearly useless. Page Up and Page Down always move the scrollbar, not the cursor. And Home and End move to the beginning and end of the document, respectively. This is inexcusable in my opinion. Every time I hit end I lose my place!

The terminal window is broken so you'll want to find a new xterm right away - GLTerm is $10 shareware and seems to work pretty well

The ports tree is missing! I never realized how convenient it was until it was gone.

AIFF was a format developed by Apple for whatever reason (I don't know). Really, nowadays, it doesn't really matter whether you prefer WAV or AIFF, since they both support the uncompressed, 44kHz 16-bit audio found on CDs.

Not to make too fine a point of it, but the X on X has delivered over a year ago already.
It's not only a plain part of XFree86 itself, with the feature-rich XDarwin tying X11 neatly into Mac OS X, including both rooted and rootless display modes. It's also complemented with an elegant window manager like OroborosX that makes the X legacy apps maximally Aqua-conformant.
There's just no reason whatsoever to waste time on Windows. None.

PS/2 Linux

I guess you have to be as old as I am to ignore the lame puns and notice the irony of IBM's role in the Open Source movement. Time was when IBM was super-proprietary -- they even had their own character code. But when mainframes went away, all of IBM exclusive tech went with it, and they became just another player in the microcomputer market. Now they're giving their tech way in order to get it accepted as a standard!

I beg to differ, there are many reasons why this can catch on:

- If it offers fast read/write/rewrite without packet writing software and just functions like a floppy, it could catch on because CD-Rw still adds another level of complexity

- If the discs are cheap (<$0.50 each) people will choose them over writeable DVD.

- You can't put writeable DVDs or CDs into your tiny digicam.

- If these discs are designed so the actual optical storage is in a very durable protective sheath, they could catch on because CDs and CD caddies are so easy to scratch/break

-Essentially, all of these add up to the 'sneakernet' factor

And to think, this could be the precursor to that little disc you see in Star Trek First Contact that Zephram Cochrane plays when he launches the warp ship. Teehee!

(it's not code it's script btw.. calling a HTML jocky a programmer is an insult to programmers everywhere)

I wouldn't describe it as a script either -- it's marked-up text, and that's all. On the other hand, writing web applications I have to do quite a bit of HTML output via templates and what not, so I guess I'm writing code to write HTML for me -- writing HTML at one remove.

As to the standards stuff -- I agree completely, I only wish more people thought it was important :/

perhaps you should just fork out some cash and buy Tenons Xtools X server for X

No, he shouldn't.

Tenon has all but abandoned Xtools. There hasn't been an update since last September, and the currently available version (1.0.4p1) is horribly unstable.

Xtools was useful for the 6-month window between the initial XF86 port to Darwin and the release of XFree 4.2 (which integrated the rootless quartz server into the main code tree). Since then, however, it's rotted. At this point, OroborosX is faster, better-featured, and much more stable.

the CLS runtime is a cheap knockoff of the JVM

And the JVM is a cheap knockoff of P-CODE.

"The intermediate code produced by the Pascal-P compiler." And this was in 1976.

i'd rather see some integration or something. Like a computer case/hamster cage.

Not a hamster, but close enough.

I'd just like to contend that Mathematica does actually handle documentation remarkably well. I've documented my entire Elementary Number Theory course using MMA including all the really really nasty formula (2^(p^(p-1.....))) in nice 2D. There is no performance loss in its display and creating them is such a breeze if you know the keyboard shortcuts (Ctrl^6 etc.). The Mathematica file format is based on Latex so anything you can do in Latex you can probably do in Mathematica too. (Well, enough anyway) I'll choose Mathematica whenever possible

Don't try serious number crunching with Mathematica. True! Tbe Runge-Kutta's and P-C methods for solving ODE's run about 100 times slower than the same in fortran. But it only took 10 lines of code so I didn't mind too much ;)

Finally, for those who want to see just how good (better IMO) Mathematica is at WYSYWIG Latex like editing, check out my ccourse notes (800kbs gzipped postscript). It has impressive formatting

Good.

Unfortunately, I don't have enough free cash to get one of these puppies. :-( However I have at least one coworker who returned his when he found out how limited they were ( and the lack of SDK ) at the time.

I hope this nod to the hackers out there is seen as a good means to motivate sales, even if only a few avail themselves of the opportunity. Between this and the PS2 Linux kit, Sony is at least starting to become a more hacker-friendly company.

Hacking hardware == good for sales.