I was thinking of the raft, indeed, and the crazy Kourier stunts, and the concert scenes. Blowing up an airport is standard stuff. I'm sure the script writer will find a way of weaseling out of the dentata. The plot doesn't really hinge on it anyway.
I think you've mistaken literature for a tech magazine. Same goes for the comment on Snow Crash, which has the added benefit of referring instead to a book that's as innovative as a fantasy novel.
There are quite a few people who don't know anything about it. They don't know what an SVM is, nor what differentiates it from linear separation (aka Perceptrons). So any explanation is more than welcome, and the GP got rightly modded up. Perhaps an even more *obvious* explanation is needed. Why don't you write one?
OO doesn't make modeling easier, it makes building GUIs and similar problems easier. I've built quite a few types of applications, from research to backoffice, and from neural network simulations to administration, and nowhere was OO useful for modeling the data. The interface yes, the rest no. Real world models seldomly show inheritance, and never multiple inheritance or dynamic message passing.
While technologically impressive, Watson is mainly association. It doesn't do much in terms of language processing; it's main strength lies in the speed with which it can search, evaluate and prune alternative associations. For psychology it's as interesting as Deep Blue.
I wish I could say the same about cognitive psychology. Rumelhart was --of course-- foremost a cognitive psychologist; AI latched on to ideas from him and likeminded colleagues. But where AI was very quick in recognizing the importance of emergent properties and subsymbolic computing and learning etc., many cognitive psychologists still think in symbolic terms and hard rules. The field consequently has hardly moved ahead since the 80s. To give an example from my own expertise: try to find a computational model of language processing that has had a significant influence on main-stream research. Initially, TRACE (by Rumelhart's direct colleagues McLelland and Elman) had some people basing their experiments on the model's outcome, but soon the theories diverted and described influences on processing in predominantly symbolic influences, such as the number of items starting with the same prefix, etc. I can only hope that cognitive psychology will pay more attention of AI...
Indeed, but we also have to recognize that HMMs and all more modern "number crunching" techniques still haven't reached human-level performance, at least not in general language processing, and do not offer anything to solve other problems, such as reasoning. Rumelhart can be credited with reviving a large field of research, and for that he deserves our respect.
No, this just leads to symptom modeling. There is no relation between "he" and "argues" or "she" and "loves" other than that they occur more frequently in the texts that comprise the corpus. I've done corpus studies, and if you look at word frequencies from a certain corpus, i.e. unigrams, they look ok, until you compare them to another one. One of them had 3rd person personal pronouns high, but the rest low, but in another, the 1st person singular (I) was the most frequent word. The difference? The former was a news paper corpus, the second e-mail.
So what did that reflect? The style and topics in the texts from which the corpus was formed, nothing else. Did you notice that "666" is associated with heaven and not hell?
So, while it is a valuable source of information for NLP, it doesn't mean anything wrt real semantics or AI. If you base your knowledge on this kind of corpus, remember the adage "garbage in, garbage out".
There are also articles on damage to the amygdala that result in loss of face recognition, about its role in emotions, in learning, etc. It seems to be just pretty important. I doubt it regulates one single thing...
In that case, it probably also correlates with the number of friends in facebook. Since correlation is causation, we finally know why the amygdala evolved: to serve Web 2.0.
It's always the same. No-one has any idea why the amygdala (or any other anatomically distinguishable brain part) should fulfill a certain function, yet we scan a few people (not too many, please!), ask them a few questions, find a correlation, and draw overreaching conclusions. It could easily be the other way around. Perhaps the amygdala has a role in face recognition (http://www.schres-journal.com/article/S0920-9964(01)00324-3/abstract), and it adapts to store all those faces in that big social network?
Neuroscience, you've gotta hate it. Glad I moved on...
Well spoken, without any hindrance of knowledge, as you gracefully admit. Field size is distributed as any other natural phenomenon: a few large, a few more mid-sized, many small. I've been working in psycho-linguistics, which is not a big field, modelling syntactic processing. I think I know all of the people that have worked in the same field the last 10 to 20 years, quite a few personally. If I get to review one of their papers, it's likely I know the author, and the topic. Of course, there are reviewers for "our" papers outside this very small group, since there are more researchers interested in these models, but I think your naivete should be cured by now.
This line of reasoning leads us nowhere. You explain "cause #1" as the result of a number of other events. They will in turn have other causes, etc., etc. This is going towards "the stuck heater got stuck because dust had accumulated; dust had accumulated because the filter was malfunctioning; the filter was malfunctioning because someone overlooked it; that person overlooked it because his telephone rang; his telephone rang because he needed to contact the office; he needed to contact the office because someone had gotten sick and he needed to do a weekend shift; the other person became sick because...". In short, nowhere, because someone getting sick doesn't cause planes to crash.
But reducing it to multiple causes doesn't mean prevention of one of them wouldn't have helped. If the analysis was correct, the absence of the trojan would have prevented the crash. So the story still stands, and it's something to consider, not to shrug off.
The parent is right. Listen to a harpischord recording at 128kbps. It's awful. Or a soprano. If you can't hear any effects of MP3 encoding, you're not trying hard enough. Perhaps 256 or 320 is fine, but Amazon doesn't sell them.
Thanks for the summary. They are really limiting preconditions, worth pointing out. Still, it's a decent achievement (that's coming from someone with a Phd in NLP). The combinatorial problem is quite huge and you need some indication that your translation makes sense. This study shows that the use of cognates and/or word structure may help. I would think it's possible to get rid of the alphabet restriction.
But these assumptions have other possible uses too. I'd think that they could be used to find relations between scripts, or showing precedence. That might clear up some historical questions.
IF anyone has a right to complain, it's the Zulus. In kwaZulu (their language), an i- is prefixed to any loan word, and the following word is then capitalized. So radio in kwaZulu would be: iRadio. Looks familiar?
How come people citing studies from the early '80s on reading normal text become experts on programming ergonomy? There are several reasons to distrust this.
1. Experience: normal readers (which were the subjects in these studies) mainly have experience in reading proportional fonts. Programmers on the other hand... 2. Context: ease of reading depends on the context, as does almost every cognitive task. Normal text does not look like code at all, so cannot be expected to be representative. 3. Importance: reading speed is not that important in programming. Reading speed for code must be way below that of normal text (which is 3 to 5 words per second for easy text, plus or minus a bit). I cannot imagine a programmer reading and understanding at that speed, so reading is not the bottleneck. 4. Errors: proportional fonts make it easier to misinterpret: 1, l or i? rn or m? 0 or O?
So, unless someone comes up with a relevant study, this should be discarded. And relevant doesn't mean speed alone.
Yes, I am a programmer and, yes, I do have a PhD and postdoc in psycholinguistics.
Come on, you're trolling! I easily memorize 34Gb per day. That's only 4 to 5 DVDs, pixel by pixel, with a few sound tracks, just over 1500 per year. No problem at all, you Thomas.
Yes, that is a perfect example. The background knowledge section uses irrelevant and obscure technical terms in the very first line (TX/RX, 2.4GHz), and goes on stating total nonsense from the layman's perspective (who wants to use it to connect to a router, not to the next farm's server), and even then doesn't refer to the term WiFi, which is all the user knows. HORRIBLE.
And any document even mentioning kernel configuration in relation to an everyday problem should be shredded, burnt, vaporized and thrown in the black hole that the LHC is going to produce Real Soon Now.
The usual definition of the term free in free market is: not under the control or in the power of another; able to act or be done as one wishes (OED). That seems to contradict the use of taxes and tariffs, although you could argue that it doesn't. Most people take the former view though.
However, the argument presented in the summary is such a load of manure, we shouldn't even bother to correct it.
No, it doesn't, part 2: if you could predict the state at any future point in time, the universe would be an FSA (Finite State Automaton/Machine), not a Turing Machine. The halting problem for FSAs is trivial.
Years ago, I worked for a natural language based search engine start-up, and the most frequent question was "sex". Yes, just three letters. Every day again. We also had a dedicated search engine for a bank, where clients could ask banking related questions. And even there people typed "sex", although it wasn't the most frequent question.
So I wonder, what did Google do to filter out the questions that I would expect most? Or did anyone ever encounter anything stronger than "make a baby"?
Slashdot Mirror
I was thinking of the raft, indeed, and the crazy Kourier stunts, and the concert scenes. Blowing up an airport is standard stuff. I'm sure the script writer will find a way of weaseling out of the dentata. The plot doesn't really hinge on it anyway.
I fear that's going to be prohibitively expensive.
I think you've mistaken literature for a tech magazine. Same goes for the comment on Snow Crash, which has the added benefit of referring instead to a book that's as innovative as a fantasy novel.
There are quite a few people who don't know anything about it. They don't know what an SVM is, nor what differentiates it from linear separation (aka Perceptrons). So any explanation is more than welcome, and the GP got rightly modded up. Perhaps an even more *obvious* explanation is needed. Why don't you write one?
And OmniGraffle does this too, but they most likely all derive their functionality from GraphViz.
OO doesn't make modeling easier, it makes building GUIs and similar problems easier. I've built quite a few types of applications, from research to backoffice, and from neural network simulations to administration, and nowhere was OO useful for modeling the data. The interface yes, the rest no. Real world models seldomly show inheritance, and never multiple inheritance or dynamic message passing.
While technologically impressive, Watson is mainly association. It doesn't do much in terms of language processing; it's main strength lies in the speed with which it can search, evaluate and prune alternative associations. For psychology it's as interesting as Deep Blue.
I wish I could say the same about cognitive psychology. Rumelhart was --of course-- foremost a cognitive psychologist; AI latched on to ideas from him and likeminded colleagues. But where AI was very quick in recognizing the importance of emergent properties and subsymbolic computing and learning etc., many cognitive psychologists still think in symbolic terms and hard rules. The field consequently has hardly moved ahead since the 80s. To give an example from my own expertise: try to find a computational model of language processing that has had a significant influence on main-stream research. Initially, TRACE (by Rumelhart's direct colleagues McLelland and Elman) had some people basing their experiments on the model's outcome, but soon the theories diverted and described influences on processing in predominantly symbolic influences, such as the number of items starting with the same prefix, etc. I can only hope that cognitive psychology will pay more attention of AI...
Indeed, but we also have to recognize that HMMs and all more modern "number crunching" techniques still haven't reached human-level performance, at least not in general language processing, and do not offer anything to solve other problems, such as reasoning. Rumelhart can be credited with reviving a large field of research, and for that he deserves our respect.
No, this just leads to symptom modeling. There is no relation between "he" and "argues" or "she" and "loves" other than that they occur more frequently in the texts that comprise the corpus. I've done corpus studies, and if you look at word frequencies from a certain corpus, i.e. unigrams, they look ok, until you compare them to another one. One of them had 3rd person personal pronouns high, but the rest low, but in another, the 1st person singular (I) was the most frequent word. The difference? The former was a news paper corpus, the second e-mail.
So what did that reflect? The style and topics in the texts from which the corpus was formed, nothing else. Did you notice that "666" is associated with heaven and not hell?
So, while it is a valuable source of information for NLP, it doesn't mean anything wrt real semantics or AI. If you base your knowledge on this kind of corpus, remember the adage "garbage in, garbage out".
There are also articles on damage to the amygdala that result in loss of face recognition, about its role in emotions, in learning, etc. It seems to be just pretty important. I doubt it regulates one single thing...
In that case, it probably also correlates with the number of friends in facebook. Since correlation is causation, we finally know why the amygdala evolved: to serve Web 2.0.
It's always the same. No-one has any idea why the amygdala (or any other anatomically distinguishable brain part) should fulfill a certain function, yet we scan a few people (not too many, please!), ask them a few questions, find a correlation, and draw overreaching conclusions. It could easily be the other way around. Perhaps the amygdala has a role in face recognition (http://www.schres-journal.com/article/S0920-9964(01)00324-3/abstract), and it adapts to store all those faces in that big social network?
Neuroscience, you've gotta hate it. Glad I moved on...
Well spoken, without any hindrance of knowledge, as you gracefully admit. Field size is distributed as any other natural phenomenon: a few large, a few more mid-sized, many small. I've been working in psycho-linguistics, which is not a big field, modelling syntactic processing. I think I know all of the people that have worked in the same field the last 10 to 20 years, quite a few personally. If I get to review one of their papers, it's likely I know the author, and the topic. Of course, there are reviewers for "our" papers outside this very small group, since there are more researchers interested in these models, but I think your naivete should be cured by now.
Nobody noticed the abbreviation for Successful Farming is SF?
This line of reasoning leads us nowhere. You explain "cause #1" as the result of a number of other events. They will in turn have other causes, etc., etc. This is going towards "the stuck heater got stuck because dust had accumulated; dust had accumulated because the filter was malfunctioning; the filter was malfunctioning because someone overlooked it; that person overlooked it because his telephone rang; his telephone rang because he needed to contact the office; he needed to contact the office because someone had gotten sick and he needed to do a weekend shift; the other person became sick because ...". In short, nowhere, because someone getting sick doesn't cause planes to crash.
But reducing it to multiple causes doesn't mean prevention of one of them wouldn't have helped. If the analysis was correct, the absence of the trojan would have prevented the crash. So the story still stands, and it's something to consider, not to shrug off.
The parent is right. Listen to a harpischord recording at 128kbps. It's awful. Or a soprano. If you can't hear any effects of MP3 encoding, you're not trying hard enough. Perhaps 256 or 320 is fine, but Amazon doesn't sell them.
Thanks for the summary. They are really limiting preconditions, worth pointing out. Still, it's a decent achievement (that's coming from someone with a Phd in NLP). The combinatorial problem is quite huge and you need some indication that your translation makes sense. This study shows that the use of cognates and/or word structure may help. I would think it's possible to get rid of the alphabet restriction.
But these assumptions have other possible uses too. I'd think that they could be used to find relations between scripts, or showing precedence. That might clear up some historical questions.
IF anyone has a right to complain, it's the Zulus. In kwaZulu (their language), an i- is prefixed to any loan word, and the following word is then capitalized. So radio in kwaZulu would be: iRadio. Looks familiar?
How come people citing studies from the early '80s on reading normal text become experts on programming ergonomy? There are several reasons to distrust this.
1. Experience: normal readers (which were the subjects in these studies) mainly have experience in reading proportional fonts. Programmers on the other hand...
2. Context: ease of reading depends on the context, as does almost every cognitive task. Normal text does not look like code at all, so cannot be expected to be representative.
3. Importance: reading speed is not that important in programming. Reading speed for code must be way below that of normal text (which is 3 to 5 words per second for easy text, plus or minus a bit). I cannot imagine a programmer reading and understanding at that speed, so reading is not the bottleneck.
4. Errors: proportional fonts make it easier to misinterpret: 1, l or i? rn or m? 0 or O?
So, unless someone comes up with a relevant study, this should be discarded. And relevant doesn't mean speed alone.
Yes, I am a programmer and, yes, I do have a PhD and postdoc in psycholinguistics.
Come on, you're trolling! I easily memorize 34Gb per day. That's only 4 to 5 DVDs, pixel by pixel, with a few sound tracks, just over 1500 per year. No problem at all, you Thomas.
Yes, that is a perfect example. The background knowledge section uses irrelevant and obscure technical terms in the very first line (TX/RX, 2.4GHz), and goes on stating total nonsense from the layman's perspective (who wants to use it to connect to a router, not to the next farm's server), and even then doesn't refer to the term WiFi, which is all the user knows. HORRIBLE.
And any document even mentioning kernel configuration in relation to an everyday problem should be shredded, burnt, vaporized and thrown in the black hole that the LHC is going to produce Real Soon Now.
The usual definition of the term free in free market is: not under the control or in the power of another; able to act or be done as one wishes (OED). That seems to contradict the use of taxes and tariffs, although you could argue that it doesn't. Most people take the former view though.
However, the argument presented in the summary is such a load of manure, we shouldn't even bother to correct it.
No, it doesn't, part 2: if you could predict the state at any future point in time, the universe would be an FSA (Finite State Automaton/Machine), not a Turing Machine. The halting problem for FSAs is trivial.
Years ago, I worked for a natural language based search engine start-up, and the most frequent question was "sex". Yes, just three letters. Every day again. We also had a dedicated search engine for a bank, where clients could ask banking related questions. And even there people typed "sex", although it wasn't the most frequent question.
So I wonder, what did Google do to filter out the questions that I would expect most? Or did anyone ever encounter anything stronger than "make a baby"?