Hmm, your post reads like a software ad for ankhor flowsheet. Interestingly, your user name "FeelYourDataFlow" is the claim on ANKHOR's web site.
So I guess your remarks on SPSS or Rapidminer, two of your direct competitors, are not really believable.
Flow-based Programming reminds me very much of current data mining suites like IBM SPSS Modeler or RapidMiner. The prefered way to model an analysis is to build a graphical flow chart using various building blocks connected to one another. Those high-level blocks can be as diverse as data transformations (filtering, aggregation) or data mining algorithms (decision tree, regression analysis).
The input for the analysis usually comes from one or many database systems. The output is usually also written to databases.
FBP makes it very easy even for non-experts to build data mining analyses using these tools.
I think, the professor in TFA is pretty much wrong.
For a reality check, please look at this Wikipedia link. It describes the current research status in the field of object recognition, a typical task in computer vision.
From the article, it should become clear that there are numerous approaches to the problem, but there is not one really capable approach that solves the problem. This is independent of computing power - we just don't know yet how to make a machine recognize non-trivial objects under real-life conditions, not matter how many CPUs it has. This fact currently limits the development of autonomic cars already.
According to Quantum_decoherence, there is no superposition of quantum states. As the radioactive element interacts with the cat and the box, its decoherence time becomes very short, so the wave function collapses almost instantly. The cat is either dead or alive, but there is no extended state of superposition.
In a situation where your company outsources code generation to some external IT consultancy, it might be beneficial for that external developer not to document his code. After all, who are they going to call when they need to understand the code or extend its functionality?
This also work down the road - if they should get several offers for future coding work in an RFP, the original programmer can still price very aggressively, as he knows all the pitfalls of the solution that he wrote in the past.
Maybe I did not make myself clear enough - I never made the assumption or the claim that the brain resembles a computer in any way.
Recognition tasks are knowledge-driven; they are based on experience. Humans have to learn to see; a major input in the visual recognition process is experience. The brain has found a way to store years of learning in its structures. So I think it is fair to say that the brain has a vast amount of storage capacity.
The aforementioned conscious calculations are an emulation. The 7-item limit would not apply, if the calculations were executed subconsciously (by whatever algorithm).
I think this line of reasoning is not convincing.
The brain masters computations (e.g. visual recognition tasks, speech recognition etc.) that require enourmous amounts of memory when implemented in a conventional computer system. Under the assumption that those recogntition tasks are inherently memory-intensive, the brain has to have similar amounts of memory at its disposal.
So, obviously, the brain is not lacking memory to execute complex calculations, but it seems to disallow the conscious control of brain structures to execute multiplications. Each single neuron could execute such multiplications with ease.
This story tries to impress us with its "IT-is-not-for-girls" spin. But honestly, you would need a lot more data to draw such a conclusion. For instance, what is the current ratio of men leaving the IT profession in general and in the middle of their career in special? There might not even be a significant difference...
Moreover, what kind of women/people are leaving the industry? As far as I know, parts of the IT market are shrinking due to Offshoring and on-going automation, so a lot of simpler IT jobs are simply vanishing right now. So the loss of those jobs might also contribute to the observation of women leaving the industry.
Unfortunately, TFA does not provide more information to put the numbers in the right context...
We are not talking about photovoltaics (i.e. the direct production of electricity from the sun), but about solar heat power plants.
The majority of power plants in this region will consist of nothing more than a whole bunch of mirrors to heat up some medium and a conventional turbine that uses the hot oil/water to generate electricity. This is a very simple technology, unlike solar panels used in photovoltaics.
Energy storage will be solved using molten salt or other liquids, but most definitely not electrical batteries. So all in all, this project is technologically very feasible. Please check http://www.desertec.org/
Well, just dumping the extended color data onto the brain might not be enough. When a child learns to see, its brain already has the basic visual perception algorithms hard-coded, e.g. there are brain structure for color detection, edge detection, motion detection etc. Those structures are built from the DNA, the genetic material, so the brain does not start learning from scratch.
Only those structures allow a child to pick up seeing as fast as it does (the process of learning to see in humans is necessary because of our ability for 3D vision. Depth perception depends on the distance between your left and your right eye; the hard-coded perception algorithms don't know this distance beforehand, so people have to learn seeing after being born. Animals without 3D vision are far more quickly able to see).
Anyway, there are no brain structures for the extended color data, so how would a brain learn the new input?
Not very likely. The slime mold in this case solve a kind of combination of a maximum flow-minimum spanning tree problem. There are algorithms with polynomial-time complexity for both problem types, so a computer would probably quickly find the optimal solution. So the computer should be way faster, even for much greater problem instances.
I don't think that's correct. According to Einstein's special theory of relativity, time passes slower for the observer - seen from the outside, it would indeed take 3000 years to fly there at light speed.
But for the traveler, time would pass much quicker, so depending on his speed, he could get there in his lifetime. There are no FTL-speeds needed here.
Not having a magnetic field is not necessarily dettrimental to the planet's atmosphere. If Earth's magnetic field failed, the atmosphere would be ionized and would self-magentize. The resulting magnetic field would deflect or redirect the solar wind.
The best example is Venus: it has an enormously dense atmosphere, yet it has no Earth-like, rotation-based magnetic field to protect it from the solar wind. What scientists have found is the effect described above: Venus has an atmospherically-created magnetic field on the sun-facing side of the planet.
Planetary atmospheres disappear because of a low planetary mass, not because of the lacking magnetic field, IMHO.
(Sources: http://adsabs.harvard.edu/abs/1979Sci...203..745R, http://adsabs.harvard.edu/abs/1984GeoRL..11..267C)
Are you familiar with the business concept of a "quick win"? In any business project, those activities will be done first to instantly show project success and to build trust with the client for the later stages. It's the same here: mankind tackles those endeavours that were easy to reach from his current standpoint. Now we might have come to a point in history, where all "quick wins" have been achieved. Nonetheless, we are still making progress (perhapes even faster than before thanks to using modern information technology), but the problems have become considerably harder.
New power sources: what about solar energy? Not exactly new, but only now close to being exploitable economically. Fusion power technology theoretically understood, but hard on an engineering level. Biotech: major advances, just think of the analysis of the DNA code. Rather effective cures for cancer available now. Life extension possbile in simpler organisms. Just watch Dr. House for an impression of the latest miracles in modern medicine.
I think it is exactly the right idea to use question-answer pairs based on common knowledge. In the time and age of SPARQL and DBpedia, this common knowledge is no longer hard to obtain. DBpedia makes it possible to ask sophisticated queries, such as as to select "people influenced by Friedrich Nietzsche" or "German musicians who were born in Berlin" formulated in SPARQL.
Now all you need to do is to build a query scheme with some exchangeable parameters like "which famous %AAA was born in %XXX in the city of %ZZZ". You put sensible values for %XXX, %ZZZ and let the SPARQL query execute. The query result will give you the answer (n.b. it does not need to be unique! It just has to exist.)
Then you use this question as your CAPTCHA, and display the right answer and some obviously wrong choices. A human will figure it out, but not a computer.
Finally, we found a use for this strange semantic web:)..cu
This is not exactly new research. The German project SIPBILD is aiming for the complete recognition of facial expressions, body language and gestures http://www.forsip.de/index.php?show=projekte_sipbild2&page=1&lang=en. The project has been running for several years now and the last publications date back from 2006.
More information on the recognition of facial expressions can be found at the TU Munich under http://vision.in.tum.de/projects/mimic_recognition/.
I am really fascinated by the research but why the hell did it take so long to figure this one out? The problem seems obvious and the solution cannot be that hard...
Hmm, thank god, the current Battlestar Galatica and Dr. Who series are new, original ideas...errh, wait...:
Battlestar Galactica first aired on September 17, 1978
Dr. Who started back in 1963
If those two series can be reanimated to be meaningful for today's generation, I am sure Knight Rider has a second life, too.
Slashdot Mirror
You made my day!! :)
Civilization IV (on Windows) is one of my guilty pleasures and surely the greatest time killer for me. It is amazingly addictive.
Hmm, your post reads like a software ad for ankhor flowsheet. Interestingly, your user name "FeelYourDataFlow" is the claim on ANKHOR's web site.
So I guess your remarks on SPSS or Rapidminer, two of your direct competitors, are not really believable.
Flow-based Programming reminds me very much of current data mining suites like IBM SPSS Modeler or RapidMiner. The prefered way to model an analysis is to build a graphical flow chart using various building blocks connected to one another. Those high-level blocks can be as diverse as data transformations (filtering, aggregation) or data mining algorithms (decision tree, regression analysis).
The input for the analysis usually comes from one or many database systems. The output is usually also written to databases. FBP makes it very easy even for non-experts to build data mining analyses using these tools.
Reminds me of GLaDOS - "There's science to be done" :)
Shufflepuck Cafe, anybody?
I think, the professor in TFA is pretty much wrong. For a reality check, please look at this Wikipedia link. It describes the current research status in the field of object recognition, a typical task in computer vision. From the article, it should become clear that there are numerous approaches to the problem, but there is not one really capable approach that solves the problem. This is independent of computing power - we just don't know yet how to make a machine recognize non-trivial objects under real-life conditions, not matter how many CPUs it has. This fact currently limits the development of autonomic cars already.
According to Quantum_decoherence, there is no superposition of quantum states. As the radioactive element interacts with the cat and the box, its decoherence time becomes very short, so the wave function collapses almost instantly. The cat is either dead or alive, but there is no extended state of superposition.
In a situation where your company outsources code generation to some external IT consultancy, it might be beneficial for that external developer not to document his code. After all, who are they going to call when they need to understand the code or extend its functionality?
This also work down the road - if they should get several offers for future coding work in an RFP, the original programmer can still price very aggressively, as he knows all the pitfalls of the solution that he wrote in the past.
"I say we take off and nuke the site from orbit. It's the only way to be sure. " - a story right out of Hollywood :)
Maybe I did not make myself clear enough - I never made the assumption or the claim that the brain resembles a computer in any way.
Recognition tasks are knowledge-driven; they are based on experience. Humans have to learn to see; a major input in the visual recognition process is experience. The brain has found a way to store years of learning in its structures. So I think it is fair to say that the brain has a vast amount of storage capacity.
The aforementioned conscious calculations are an emulation. The 7-item limit would not apply, if the calculations were executed subconsciously (by whatever algorithm).
I think this line of reasoning is not convincing.
The brain masters computations (e.g. visual recognition tasks, speech recognition etc.) that require enourmous amounts of memory when implemented in a conventional computer system. Under the assumption that those recogntition tasks are inherently memory-intensive, the brain has to have similar amounts of memory at its disposal.
So, obviously, the brain is not lacking memory to execute complex calculations, but it seems to disallow the conscious control of brain structures to execute multiplications. Each single neuron could execute such multiplications with ease.
This story tries to impress us with its "IT-is-not-for-girls" spin. But honestly, you would need a lot more data to draw such a conclusion. For instance, what is the current ratio of men leaving the IT profession in general and in the middle of their career in special? There might not even be a significant difference... /people are leaving the industry? As far as I know, parts of the IT market are shrinking due to Offshoring and on-going automation, so a lot of simpler IT jobs are simply vanishing right now. So the loss of those jobs might also contribute to the observation of women leaving the industry.
Moreover, what kind of women
Unfortunately, TFA does not provide more information to put the numbers in the right context...
We are not talking about photovoltaics (i.e. the direct production of electricity from the sun), but about solar heat power plants.
The majority of power plants in this region will consist of nothing more than a whole bunch of mirrors to heat up some medium and a conventional turbine that uses the hot oil/water to generate electricity. This is a very simple technology, unlike solar panels used in photovoltaics.
Energy storage will be solved using molten salt or other liquids, but most definitely not electrical batteries. So all in all, this project is technologically very feasible. Please check http://www.desertec.org/
I can't help it but those pictures look like straight out of a Hollywood science fiction movie...this is just awesome! We are living in the future :)
Well, just dumping the extended color data onto the brain might not be enough. When a child learns to see, its brain already has the basic visual perception algorithms hard-coded, e.g. there are brain structure for color detection, edge detection, motion detection etc. Those structures are built from the DNA, the genetic material, so the brain does not start learning from scratch.
Only those structures allow a child to pick up seeing as fast as it does (the process of learning to see in humans is necessary because of our ability for 3D vision. Depth perception depends on the distance between your left and your right eye; the hard-coded perception algorithms don't know this distance beforehand, so people have to learn seeing after being born. Animals without 3D vision are far more quickly able to see).
Anyway, there are no brain structures for the extended color data, so how would a brain learn the new input?
Not very likely. The slime mold in this case solve a kind of combination of a maximum flow-minimum spanning tree problem. There are algorithms with polynomial-time complexity for both problem types, so a computer would probably quickly find the optimal solution. So the computer should be way faster, even for much greater problem instances.
I don't think that's correct. According to Einstein's special theory of relativity, time passes slower for the observer - seen from the outside, it would indeed take 3000 years to fly there at light speed.
But for the traveler, time would pass much quicker, so depending on his speed, he could get there in his lifetime. There are no FTL-speeds needed here.
Not having a magnetic field is not necessarily dettrimental to the planet's atmosphere. If Earth's magnetic field failed, the atmosphere would be ionized and would self-magentize. The resulting magnetic field would deflect or redirect the solar wind.
The best example is Venus: it has an enormously dense atmosphere, yet it has no Earth-like, rotation-based magnetic field to protect it from the solar wind. What scientists have found is the effect described above: Venus has an atmospherically-created magnetic field on the sun-facing side of the planet.
Planetary atmospheres disappear because of a low planetary mass, not because of the lacking magnetic field, IMHO.
(Sources: http://adsabs.harvard.edu/abs/1979Sci...203..745R, http://adsabs.harvard.edu/abs/1984GeoRL..11..267C)
Are you familiar with the business concept of a "quick win"? In any business project, those activities will be done first to instantly show project success and to build trust with the client for the later stages.
It's the same here: mankind tackles those endeavours that were easy to reach from his current standpoint. Now we might have come to a point in history, where all "quick wins" have been achieved. Nonetheless, we are still making progress (perhapes even faster than before thanks to using modern information technology), but the problems have become considerably harder.
New power sources: what about solar energy? Not exactly new, but only now close to being exploitable economically. Fusion power technology theoretically understood, but hard on an engineering level.
Biotech: major advances, just think of the analysis of the DNA code. Rather effective cures for cancer available now. Life extension possbile in simpler organisms. Just watch Dr. House for an impression of the latest miracles in modern medicine.
I think we are doing fine in this respect.
I think it is exactly the right idea to use question-answer pairs based on common knowledge. In the time and age of SPARQL and DBpedia, this common knowledge is no longer hard to obtain. DBpedia makes it possible to ask sophisticated queries, such as as to select "people influenced by Friedrich Nietzsche" or "German musicians who were born in Berlin" formulated in SPARQL.
:) ..cu
Now all you need to do is to build a query scheme with some exchangeable parameters like "which famous %AAA was born in %XXX in the city of %ZZZ". You put sensible values for %XXX, %ZZZ and let the SPARQL query execute. The query result will give you the answer (n.b. it does not need to be unique! It just has to exist.)
Then you use this question as your CAPTCHA, and display the right answer and some obviously wrong choices. A human will figure it out, but not a computer.
Finally, we found a use for this strange semantic web
This is not exactly new research. The German project SIPBILD is aiming for the complete recognition of facial expressions, body language and gestures http://www.forsip.de/index.php?show=projekte_sipbild2&page=1&lang=en.
The project has been running for several years now and the last publications date back from 2006.
More information on the recognition of facial expressions can be found at the TU Munich under http://vision.in.tum.de/projects/mimic_recognition/.
I am really fascinated by the research but why the hell did it take so long to figure this one out? The problem seems obvious and the solution cannot be that hard...
Hmm, thank god, the current Battlestar Galatica and Dr. Who series are new, original ideas...errh, wait...: Battlestar Galactica first aired on September 17, 1978 Dr. Who started back in 1963 If those two series can be reanimated to be meaningful for today's generation, I am sure Knight Rider has a second life, too.