Slashdot Mirror

(where are those sarcasm tags again?)

Well:
1) If you measure all the sources of radiative forcing, you see that the natural ones are pretty much negligible with respect to the current warming, where as the "human-caused" ones are large.

2) There have been papers that split the warming into the warming that would have happened from natural forcing, and that which would have happened from anthropogenic forcing. ((paper). Satisfyingly, the warming that has happened from the sum of the forcings, is approximately the sum of the warmings from each forcing. So it's nice and additive, therefore statements like "x% of the warming of the past y years is anthropogenic" are meaningful. Such as "80% of the warming of the past 100 years is anthropogenic" or "110% of the warming of the past 50 years is anthropogenic".

And here are their research centers which are analogous to institutes.

EPFL stands for Ecole Polytechnique Fédérale de Lausanne (Lausanne Federal Polytechnic School). It is not a "research institute" but one of the two (with ETH Zürich) swiss federal engineering schools, and very renowned at that. Lots of good research labs there, too, just like at MIT or Caltech, but that doesn't make them "research institutes" neither.

Its mission may not be pure research, but it implicitly self-identifies as a research institute (among other things) on its website:

With more than 300 laboratories and research groups on campus, EPFL is one of the Europe's most innovative and productive technology institutes. At EPFL the emphasis is on both theoretical and applied research.

How about converting the binary directly?

X86->LLVM IR->anything:
http://infoscience.epfl.ch/rec...

Opensource, too. repository:
https://dslabgit.epfl.ch/git/s...
(checkout revgen)

has anyone tried it?

How about converting the binary directly?

X86->LLVM IR->anything:
http://infoscience.epfl.ch/rec...

Opensource, too. repository:
https://dslabgit.epfl.ch/git/s...
(checkout revgen)

has anyone tried it?

actual referenced article, not lousy BusinessInsider. http://moodle.epfl.ch/pluginfi... Roadmap.pdf

IBM's blue brain project has been simulating real brains by painstakingly mapping slices of rat/cat brains onto their software model for more than a decade now. IBM's "Watson" appears to me to be an spin off from that project. The Jeopardy "stunt" proved Watson is indisputably superior to the best humans at general knowledge questions (an open ended problem domain). IBM have developed similar 'brain on a chip" technology and have been using it for a while now. The hardware that won the Jeopardy games a couple of years ago required "20 tons of equipment", IBM are just now starting to lease instances of Watson to "development partners", last time I checked each Watson clone runs on a 50kg "bar fridge" server.

The Blue brain project is primarily aimed at medical research and I believe it's now part of the EU's larger and more ambitious Human brain project.

Here it comes: Hate to say "I told you so" but..... I said "it's here" when I saw the Jeopardy stunt, my SO looked at me and said "It's looking up the answers, what's the big deal?". The "big deal" of course is that it finds the correct answer from the mass of unstructured textual data returned by a simple web crawler, which from a black box POV strongly implies it "understands" the question. Further, when Watson falls for bullshit he reads on the internet or lacks context, the developers correct the misunderstanding by "teaching it" new facts. As one of the developers puts it, it the computer knows everything there is to know about human anatomy how do you get it to correctly interpret the phrase "Noses run and feet smell"? Systems like Watson can do this themselves, essentially they are finding meaning by reading unstructured data! Which at the end of that day is conceptually no different to what any other "brain" does.

prophecy/
Anyone who wants a software dev job in the not too distant future better start thinking less about programming them and more about training them.
/prophecy

Here is an open conference scheduled next week by Ari Juels in Lausanne, CH:

The event
 

HOLY SHIT! Exactly how hard are these things bouncing off of stuff? They could just as well go through the walls!

If you are a software developer, and don't find use for arbitrarily large number of cores... Time to get up to date!

That's why I'm so excited by the new breed of languages like Scala.

Sure, there's no silver bullet to automagically solve all parallel programming problems, but languages like Scala have features like Parallel Collections libraries, functional programming and Parallel Domain Specific Languages that can abstract enough of the problems of parallel programming away that journeyman programmers have a decent chance of being able to work effectively with multiple cores.

I'm somewhat disappointed by the adoption curve. The reluctance to move toward an actual solution to the problem is somewhat surprising.

This is also the real reason behind NoSQL databases... the need to scale horizontally instead of vertically is the primary driver, not a disdain for SQL.

one small step for a university, one huge leap for our roadmap towards simulating a brain. Another one recent example of our progress in this was the Spaun brain model ( a small one that is, IIRC 12million neurons ) which was featured on slashdot as well, and also the older blue brain project http://bluebrain.epfl.ch/

I can't wait for the moment ( within 20 years hopefully ) when we will have a full human brain simulation. the possibilities from that point are endless. Maybe our last invention!

your "assumptions" and "i will be impressed if..." are overly pessimistic. ever heard of http://bluebrain.epfl.ch/ ?

although multicores are good, there just aren't that many decent parallel programmers out there. I (and a few others) find parallel programming easy

That's why languages like Scala are so appealing.

Sure, there's no silver bullet to automagically solve all parallel programming problems, but languages like Scala have features like Parallel Collections libraries, functional programming and Parallel Domain Specific Languages that can abstract enough of the problems of parallel programming away that journeyman programmers have a decent chance of being able to work effectively with multiple cores.

Parent link is bad. Try this:

http://lmts.epfl.ch/microthrust

This ion thruster is unique by its extremely small size. we have miniaturized not only the ion emitter, but the entire thruster including high-voltage electronics and tank. Our complete thruster has a mass of 200g (including 100 ml of fuel), thus allowing it to be used on nanosatellites. It is the first high efficiency electric propulsion system that can be used in cubesats and 5kg satellites, such as those being planned for OLFAR The principle of operation of colloid thruster a bit different from the ion engines used fro instance on SMART-1, which uses ionize Xenon. in our case, we use a particular conductive liquid, an ionic liquid, from which we can extract both positive and negative ions. using a liquid avoids a pressurized tank, and allows for important simplification of the system (no valves, no heavy tanks, all flow controlled by capillary and electrostatic forces. using the ionic liquid allows the same speed as using a gas, but offers one big advantage: since we emit (from 2 chips in parallel) both positive and negative ions, the spacecraft stays electrically neutral, which is essential for electric propulsion to avoid having the ions fly back to the spacecraft. for more conventional electric propulsion systems, only positive ions can be emitted, so a neutralizer is needed to emit electrons to keep the spacecraft charge neutral. not having a neutralizer allows significant mass and power savings.
I'm biased, 'cause I work on this!
http://lmts.epfl.ch/microthust
- Herb Shea

The propulsion system emits ions at high speed (40 km/s) and is thus very efficient at converting propellant mass to satellite momentum. Thrust is low, but given time, ver lge orbit chanegs are possible. for example, in order to reach lunar orbit from low-Earth orbit, a 3-kg nanosatellite with our motor would travel for about 2 years and consume about 500 grams of fuel" - Herb Shea

The propulsion system emits ions at high speed (40 km/s) and is thus very efficient at converting propellant mass to satellite momentum. Thrust is low, but given time, ver lge orbit chanegs are possible. for example, in order to reach lunar orbit from low-Earth orbit, a 3-kg nanosatellite with our motor would travel for about 2 years and consume about 500 grams of fuel" - Herb Shea

The link you provide supports that this is selection bias - he cracked 26025 out of 93688 passwords, and then made the brilliant deduction that boils down to "of those passwords that I easily cracked, most were found to be easily cracked". No shit, Sherlock.

I didn't say that the link disproves that the selection bias exists---it simply doesn't exhibit that selection bias, because it represents a sample of all passwords used on a site. The top 30 passwords were not "the top 30 that were cracked"; they were the top 30 passwords used on the site at all. I could determine this because they were stored as unsalted MD5 hashes.

Sure, that 36% of passwords are easily cracked is bad in itself, but that's another thing entirely. It can't be used as statistics to extrapolate anything using the word "most". It only applies to that subset of weak password.

Yes, I don't dispute that.

I also have to arrest you for " I found that 36% of all passwords were easily discoverable using a rainbow table". This is incorrect. 100% of all passwords are easily discoverable using a rainbow table. 36% may be easily discoverable using a partial rainbow table, which is not the same thing.

What is the difference between a rainbow table and a "partial rainbow table", in your view? Do you think a "rainbow table" means a table containing all possible passwords? Considering that many hash functions have an infinite number of possible inputs, there's no such thing as a rainbow table, according to that definition. It also seems to contradict the usage by Philippe Oechslin in the paper which introduced the term.

Incidentally, this is how "break" is implemented in Scala. There is no break keyword in the language, but it is implemented at the library level by throwing/catching an exception.

Here's the source code:
https://lampsvn.epfl.ch/trac/scala/browser/scala/tags/R_2_9_1_final/src//library/scala/util/control/Breaks.scala#L1

Every...study has shown that the best available VP8 encoders require almost 2x the bitrate of the best H.264 high profile encoders.

Got links?

Sure... this is the most comprehensive qualitative test I've seen, using a huge varietry of sources and metrics. See conclusions section on page 93, which shows WebM requiring >2x the bits of x264 for the same quality.

A rigorous subjective comparison can be found here, using the Double Stimulus Continuous Quality Scale methdololgy.

Note in both subjective and objective comparisons, WebM takes 2x or more bits to achieve the same quality at web bitrates of ~500 kbps.

At much higher bitrates, the quality differences narrow. But high bitrates aren't valuable for Internet use cases, and in any case at 2.5 Mbps and SD resolution, even inefficient codecs like MPEG-2 or WMV8 look good.

FYI, there is a new attack on TKIP that can recover the temporal key after capturing 2^38 packets (for comparison, WEP's IV is 24-bit):
http://infoscience.epfl.ch/record/165984

Who works on Artificial Brain then? Don't say AI researchers.

These guys

http://slideshot.epfl.ch/play/ktn_katerina

no idea why u got modded flamebait :)

If you are curious about the slightly deeper and murkier details, this will tell you why handling DDoS attacks is still difficult.

We have to be patient with the big dinosaurs. Word travels slowly inside such a large company.

1 June 2001, Ballmer's legendary comment, "Linux is a cancer that attaches itself in an intellectual property sense to everything it touches. That's the way that the license works."

From the comments in TFA, 8 April 2002 - FUD from Juan Gonzalez, General Manager of Microsoft Peru gets shot down in flames by Congressman Edgar David Villanueva Nuñez

20 February 2003, David Stutz, retiring group program manager, "delivered a kick in the pants to his former employer" saying "Microsoft is in danger of being swept away by open source"

12 May 2004, Windows Template Library (and Windows Installer XML) posted to sourceforge. The blogosphere reels in shock. Even slashdot isn't sure what to think.

I got bored at this point, but there's lots more popcorn-hour fun and games as this large corporation tries to deal with a rapidly changing industry.

The Official Solar Impulse website at epfl

The Official Solar Impulse website at epfl

I've never had Tetris dreams, but after spending 80 hours over spring break doing VLSI (what was I thinking), I did dream about being attacked by colorful rectangular geometry and having to make it all fit together to make it stop. Does that count?

"You're advocating the "emergent intelligence" model of AI, where intelligence "somehow" is created by the confluence of lots of data...[snip]...In practice the degrees of freedom which unstructured data provides far exceed the capability of current (and likely future) computers."

You sure about that?. They have already created a molecular level model of the mammalian neocortex and the expected date for completion of a full model of the mammalian brain is solely dependent on the amount of money thrown at it. The model neocortex can already faithfully recreate patterns seen in fMRI scans. If given the first part of a pattern it will acurately reproduce the rest of it. The project is mainly geared toward medicine but they have also inserted the model into an artifical world in order to study it's capacity for learning.

Depending on your choice of religion, the structure is the result of divine intervention or millions of years of chance and evolution. When building AI systems, the problem has always been to find the appropriate structure or features.

The dualisim of Descates has been thouroughly debunked and I'm sure you are aware that evolution is not a religion. The mind does "somehow" emerge from the brain's deterministic processing of a continuous avalanche of unstructured data. Looking for the structure of mind is like looking for the structure of fog from within the fog bank. This is why it's called the hard problem of conciousness, the mistake most people make is that we need to solve that problem before we can create an artificial mind. After all the pyramids were built with levers long before the greeks came along and explained why a lever "somehow" inreases the power of the person using it.

The real question is will we recognise an artificial mind if one emerges from an artificial brain. It's unlikely that such a mind would pass the turing test but we already have lots of examples of minds in our mammalian cousins that are also unable to pass the turing test.

3 isn't really applicable as not only is it optimised for it's own version of java, but you have the option of writing C anyway.

GP's point #3 was about C++ being more efficient by a lot. His point actually is valid. Consider all the things you need to be mindful of when writing efficient Android apps in Java There are so manu things that you need to be careful not to trip over, even in Android's optimized Java. Things that a C++ compiler would optimize out no problem. Generally Android's java is very optimized and is terrific and surprisingly fast, but it's still no C++ in terms of performance. But it's also easy to write for generally and very approachable, so I'm not complaining. Just saying that Gp's point #3 should not be dismissed entirely without due consideration.

What about these guys? They've been mentioned by Slashdot before. Basically they're trying to completely simulate a complete (rat's) neocortical column, a basic building block of the cerebral cortex. Now, I know that IBM has oodles of money and resources, but if TFA is to be believed, IBM is saying they did 3 years ago what the BlueBrain researchers are saying is 10+ years away (complete simulation of a rat's cortex, consisting of hundreds of thousands of neocortical columns). Either IBM is making scarily fast advancements in this field, or BlueBrain is doing this the hard way.

Scala is great, but one really annoying thing about it is that it inherits type erasure implementation of generics from Java. This means that you cannot overload methods on argument with the same generic class with different type parameters, cannot implement the same generic interface with different type parameters on the same class, cannot check whether a class implements a particular generic interface for a given type parameter, etc. They did fix some issues - for example, you can instantiate arrays - but it's still far from perfect.

I understand the need to match Java's broken model for the sake of interoperability, but surely a better way can be devised for pure Scala code? It's pretty much the only area where Scala noticeably lags behind advanced .NET-hosted languages (such as Nemerle or F#).

"It's not a hardware breakthrough that'll create a true AI - it's an algorithm breakthrough that's required."

On the contray, I think you need a algorithmic breakthrough to understand the brain but you don't need a new algorithim to create a brain. Humans have built and used many things well before they had a theoretical basis for how they worked, for example people were using levers to build pyramids long before archimedes came and gave us the "lever algorithim".

Please read:

Making a Faster Cryptanalytic Time-Memory Trade-Oï
Philippe Oechslin
http://lasecwww.epfl.ch/~oechslin/publications/crypto03.pdf

or any other paper in its references

"And they only need to increase that by 100,000 times to get to about the same number of neurons as a human brain, let alone the synaptic connections (which would be somewhere on the order of 2,000,000 times what they've done)."

Not as far fetched as it once seemed.

From the link: "At the end of 2006, the Blue Brain project had created a model of the basic functional unit of the brain, the neocortical column. At the push of a button, the model could reconstruct biologically accurate neurons based on detailed experimental data, and automatically connect them in a biological manner, a task that involves positioning around 30 million synapses in precise 3D locations."

Note that some major parts of the model are down at the molecular level. Since then experiments using data from brain scans have shown that the simulated neocortex appears to behave like a real one.

I doubt people (particularly the religious) will accept a computer consciousness. A good number of scientists belive animals are prue programming (nobody home just trainable automata) and there are a shitload of ordinary people out there who still don't belive climate simulations are usefull predictors (scroll down to embedded movie).

Okay, I guess I'll do it for you.

SLAM

BLAST

yogi

pldi

popl

Well, Microsoft researchers are involved, to some extent, in some research that is, well, extending some old stuff in ways that might be new. Groundbreaking, maybe, to some people.

Maybe these tools will help generate "correct" code for some definition of correctness. But have these guys defended their choice of definition of "correctness"? Have they shown how it applies to the real world? Is the application field a niche field, or will it help with OSses and general end-user applications?
But, to me, it just seems to be heading the wrong direction. I've been there. All I could find down those paths is more of the same blind alleys. Maybe they'll find something interesting, if so, good for them.

Does it really help solve the problems were are facing in the current market? How does it help users solve their junk e-mail problems? How does it clean up the botnets? How does it prevent users from clicking OK and adding to the botfarms?

How does it give users safe, secure, _minimal_ browsers for checking their bank accounts and making payments for purchases?

Research is all well and good, but this is not what the market needs now.

Odersky's work on Scala goes back quite a few years, so its not like Scala is a fly by night operation.

He did not call it Scala back then but designing a FPish language on the JVM is something he has been doing for quite a while. Some of the older work they did formed the basis of Java Generics and that was done back in 1997, when Java was only 3 years old.

http://people.epfl.ch/martin.odersky

Ruby was used in Japan for many years before becoming more widely known in the West (at a time the main webpage was in Japanese). Ruby is OLDER than Python and about as old as PHP and Java.

Ruby was stagnating though, version 1.8, the current common one, was released quite a few years ago.

Personally, I have been using it as a Perl/Pythong general shell scripting language since about 2000.
 

Scala also keeps Java's strong static typing and adds functional language features. I don't think it needs any development at all to be adapted for mainstream use.

Scala is a great thing, now what it needs is equally great tooling (i.e. IDE support, including refactoring, on the same level as we have for Java). And it's getting there - there is an Eclipse plugin, and a NetBeans one - but it definitely needs more work and polish.

The payoff would be huge, though. Right now, as far as languages go, C# is far more advanced than Java. But Scala is equally more advanced than C# (the only thing on .NET that could compare with Scala is F#, and that's less stable and mature currently... but it will stabilize once VS2010 is released, so there isn't much time). Some significant investment into Scala by one of the big players could help straighten things out a bit (it's good when there are two major competing languages in the same niche, because it forces them to evolve at a more rapid pace, steal each other's features, and learn from each other's mistakes - as Java and C# did for several years, until Java began to stagnate).

About the only thing I don't like about Scala is generic type erasure it inherited from Java to maintain class compatibility; but there are workarounds already, and if Scala becomes The Next Language for the Java platform, it may well deprecate erasure and introduce its own reified generics then.

Now, if only someone would pick it up and throw money (and marketing resources) at it. Like, say, Google...

Yep that was Lorenz. I assume the brain map in TFA will be useful to the Blue brain project

Read this and all will become clear:
Event-Based Programming without Inversion of Control

A typical neuron is a vastly complex electro-chemical computer,

You can still simulate these interactions digitally and have the output match. Like these guys did.

Yeah and we all know who the employers" are....no..no,...not the fucking cheese video....argggggg!!!!!

it has some support for using version control (using some version control software called RCS). I haven't tried it yet, but I've been tempted.

Trust me, there is nothing tempting about using RCS.

I use LyX to write my LaTeX docs, and it has some support for using version control (using some version control software called RCS). I haven't tried it yet, but I've been tempted.

Thus far, I've been in the position where I just write most of my contribution in Lyx, then export it to plain Latex and sent it to collaborators. From there we just do the collaboration in plain Latex. The problem for me hasn't been the lack of version control but rather the ability/willingness of collaborators to all use LyX. Now, one can import LaTeX into Lyx, but if you do a closed loop (write -> export -> import again) you'll find things are not quite as nice in the end, so this hasn't seemed to be an optimal solution.

As for people saying that technical writers ought to be able to use technical software: A) in many cases it's a question of willingness to commit the time, not ability and B) just because you're technically knowledgeable in, say, cosmological physics, doesn't mean you're adept with computers. ...trust me on this one.

The demonstration given by Martin Vuagnoux and Sylvain Pasini from the LASEC/EPFL has already been slashdotted (see http://hardware.slashdot.org/article.pl?sid=08/10/20/1248234&from=rss) in october 2008. You can see the videos of the experiment on http://lasecwww.epfl.ch/keyboard

And me. I remember the videos: http://lasecwww.epfl.ch/keyboard/ Engadget reported it last October: http://www.engadget.com/2008/10/20/keyboard-eavesdropping-just-got-way-easier-thanks-to-electrom/

Even my crappiest stereo3d rig was 1600x1200.

I'm assuming you are referring to a monitor/shutter glasses type system. That's an inaccurate comparison with a head-mounted display. You have to remember that the screen is < 2in. away from your eyes. Most of the entry-level 'pro' stereo head-mounted displays ($5000+) don't get over 1024x768, so $500 for 640x480 really isn't that bad.

I think the motion tracking adds more to the experience than absolute resolution. Your mind can fill in the visual blanks with a little imagination. It can't make up for the lack of basic orientation tracking, parallax effects, and changes in sound with head position. You can add basic head tracking/parallax to a crt/glasses system but the tracking is heavily constrained by the user having to face the monitor. If you want to look skywards you have to adjust your avatar's view with the mouse/keyboard vs just looking up to see above you. This constrained tracking really isn't useful for anything more than simple parallax effects.

My interest in head-mounted displays is for exploring 3d data visualizations and for performing experiments. I can see how 640x480 might be a disappointment for a gamer, but it would work just fine for my uses... Although 640x480 sounds really low it doesn't look that bad. If you render at a higher resolution and scale/filter to 640x480 the results can look quite nice

at the Laboratory for cryptologic algorithms.

Here you are: http://gnuwin.epfl.ch/articles/en/reponseperou/villanueva_to_ms.html The original was in Spanish, it's been translated into many languages. Brilliant - but Microsoft still won, if I recall correctly.