1317 messages over 165 pages: << Previous 1 2 3 4 5 6 7 ... 117 ... 164 165 Next >>
Jeffers Senior Member United Kingdom Joined 4918 days ago 2151 posts - 3960 votes Speaks: English* Studies: Hindi, Ancient Greek, French, Sanskrit, German
| Message 929 of 1317 31 January 2014 at 4:56pm | IP Logged |
geoffw wrote:
Forgive my ignorance, but isn't "Parallel Distributed Processing"
redundant? Is it
possible to do parallel processing without distributing the code and data? |
|
|
It isn't possible to do parallel processing which isn't distributed, but it is
possible to do distributed processing which isn't parallel. So, the term isn't
completely redundant.
1 person has voted this message useful
| geoffw Triglot Senior Member United States Joined 4697 days ago 1134 posts - 1865 votes Speaks: English*, German, Yiddish Studies: Modern Hebrew, French, Dutch, Italian, Russian
| Message 930 of 1317 31 January 2014 at 4:57pm | IP Logged |
Jeffers wrote:
It isn't possible to do parallel processing which isn't distributed, but it is
possible to do distributed processing which isn't parallel. So, the term isn't
completely redundant. |
|
|
The possibility of distributed, non-parallel processing occurred to me before posting,
but I was pretty sure that wasn't the distinction being made.
1 person has voted this message useful
| patrickwilken Senior Member Germany radiant-flux.net Joined 4542 days ago 1546 posts - 3200 votes Studies: German
| Message 931 of 1317 31 January 2014 at 4:59pm | IP Logged |
geoffw wrote:
Sounds like you mean that "distributed" means that the data (semantic meaning, symbols)
is distributed, as opposed to merely distributing processing functions (e.g., optimized
code may execute different sections of a program in parallel, but each processing node
handles all the data for that particular section). |
|
|
I am sure Eric an answer this better.
But yes, the "distributed" is referring to the semantics/symbols. In fact, I think they used to talk about 'sub-symbolic' processing. The idea being that there weren't symbols in the computation as such.
The two PDP books were really the breakthrough volumes that argued against traditional AI symbolic approaches, and paved the way for modern neural network approaches in the study of cognition.
Of course, now days none of this is seen as such a big deal, but at the time I remember it being extremely exciting and a radical attack on prevailing approaches to cognition. It was the area all the cool kids wanted to be involved in.
Edited by patrickwilken on 31 January 2014 at 5:07pm
2 persons have voted this message useful
|
emk Diglot Moderator United States Joined 5541 days ago 2615 posts - 8806 votes Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map
| Message 932 of 1317 01 February 2014 at 1:33am | IP Logged |
patrickwilken wrote:
emk wrote:
My problem with NGE is, well, the way that all of the women look like blow-up dolls, and the way that the camera is utterly obsessed with their bodies. It's distracting, and it feels like the show is trying way too hard to manipulate the viewer. |
|
|
It's been ages since I saw the show, but I remember thinking that there was a lot of depth to all the main characters in the show, and the female characters were presented as strong and interesting individuals. |
|
|
Well, I'm pleased to say that the last two episodes were a lot better—they focused on the characters, and not so much on gratuitous Barbie-doll closeups. For anybody else who's tempted by the series, give it few episodes. It starts slow and it takes a while to introduce the characters.
I wound up leaving the French subs on for about 3 episodes and then turning them off. For some reason it took a while to "tune in" on the dialog.
patrickwilken wrote:
The two PDP books were really the breakthrough volumes that argued against traditional AI symbolic approaches, and paved the way for modern neural network approaches in the study of cognition. |
|
|
This whole digression about AI is getting a little off-topic, but it does influence how I approach language learning. So one more post, and then I'll move on. :-)
When I was younger, I actually worked with some of the old symbolic AI folks. The 1980s really were an AI goldrush. For a while, there was a lot of money flying around, and there were some ferociously smart people working on these problems. Unfortunately, nobody succeeded in delivering useful results, and the various corporate and military sponsors stopped spending money. This, in turn, led to the "AI winter" and a lot of bankrupt Lisp companies.
Why did symbolic reasoning fail? Well, I could talk about this for hours. But other people have said it better:
The Semantic Web, Syllogism, and Worldview: Clay Shirky explains why syllogisms will never do anything useful.
On Chomsky and the Two Cultures of Statistical Learning. Peter Norvig (Google's Director of Research and the author of a fine AI textbook) explains why understanding language is an inherently statistical endeavor.
Given the hours I've spent working alongside really smart people who were once heavily invested in symbolic AI, I want to say that Norvig's claims all make excellent sense, and Chomsky seems to be firmly trapped in the same tarpit that swallowed all the first-generation AI companies.
What does all this have to do with language learning? Well, among other things, it strongly suggests that you can't internalize a language simply by memorizing lots of grammar rules and lots of exceptions, and then adding 10,000 L1<->L2 vocabulary cards in Anki. If language were actually that simple, computers would already be able to talk.
Language is full of deep statistical patterns, and these patterns can only be internalized via massive exposure. You can either dive straight into those patterns and occasionally tweak your grammar later (which is how I learned French), or you can learn some grammar up front and use it as a tool to manually decode lots of input (I'm having to do a lot more of this with Egyptian, thanks to the deplorable lack of native TV shows). But in either case, there's no way to get good at a language simply by learning rules: You've ultimately got to put your brain in contact with the language, and hope that some magic happens. Studying is useful, but it's only a small part of the overall process.
Even Google Translate actually needs to slurp up half the web and try to extract statistical patterns from an ocean of input. Whether human or machine, there's no way around this.
6 persons have voted this message useful
|
emk Diglot Moderator United States Joined 5541 days ago 2615 posts - 8806 votes Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map
| Message 933 of 1317 01 February 2014 at 1:12pm | IP Logged |
Yay! My new iTalki "community" tutor rules. He reads a lot of science fiction, and he's happy to just spend an hour talking about stuff. Last night, we discussed:
- Ender's Game and Orson Scott Card's increasingly extreme politics,
- varieties of American Protestantism,
- why Bernard Werber's French science fiction is complètement nul (and why I should avoid Marc Levy and Musso),
- Le Trône de fer (both the books and the TV series), and why it's fundamentally a depressing series (but still fun!),
- Barjavel's La Nuit des temps, plus Le Déchronologue and La Horde du Contrevent,
- and all kinds of other stuff.
This is pretty much the conversation that I need to be having in French right now. And I need to be having it without relying on the safety-net of matrimonial telepathy. In my personal experience, matrimonial telepathy is a major obstacle to progress after B2.
During these sessions, I'm not asking for corrections at all. I'm working on speed and fluency, and if I make a few grammar mistakes, I'll either correct them quickly or push on. And the other nice thing about iTalki is (1) I get to use Skype, not some nasty web-based system, and (2) there are some great, inexpensive tutors if all I want is a challenging conversation.
2 persons have voted this message useful
| Jeffers Senior Member United Kingdom Joined 4918 days ago 2151 posts - 3960 votes Speaks: English* Studies: Hindi, Ancient Greek, French, Sanskrit, German
| Message 934 of 1317 01 February 2014 at 1:55pm | IP Logged |
I decided to look into Neon Genesis Evangelion, and just wanted to check what you've gotten, emk.
A search on amazon.co.uk came up with single disc editions costing about £16, and with only English and Japanese audio. Amazon.fr has a 7 disc edition for about 16 euros, and is supposed to have 26 episodes. The amazon.fr version only has French and Japanese (and French subtitles).
Does either of these match what you got? Do the French subtitles come close to the French audio? The amazon.fr edition is the obvious choice, but I wish there was an English track so I could buy it for my son as a gift! Sneaky way to get myself more language resources, heh heh heh.
1 person has voted this message useful
| patrickwilken Senior Member Germany radiant-flux.net Joined 4542 days ago 1546 posts - 3200 votes Studies: German
| Message 935 of 1317 01 February 2014 at 2:04pm | IP Logged |
emk wrote:
Well, I'm pleased to say that the last two episodes were a lot better—they focused on the characters, and not so much on gratuitous Barbie-doll closeups. For anybody else who's tempted by the series, give it few episodes. It starts slow and it takes a while to introduce the characters.
|
|
|
That's great. It did occur to me that the dubbing might actually account for some of the perceived sexism too - depending on how the voice-overs work.
emk wrote:
What does all this have to do with language learning? Well, among other things, it strongly suggests that you can't internalize a language simply by memorizing lots of grammar rules and lots of exceptions, and then adding 10,000 L1<->L2 vocabulary cards in Anki. If language were actually that simple, computers would already be able to talk.
Language is full of deep statistical patterns, and these patterns can only be internalized via massive exposure. You can either dive straight into those patterns and occasionally tweak your grammar later (which is how I learned French), or you can learn some grammar up front and use it as a tool to manually decode lots of input (I'm having to do a lot more of this with Egyptian, thanks to the deplorable lack of native TV shows). But in either case, there's no way to get good at a language simply by learning rules: You've ultimately got to put your brain in contact with the language, and hope that some magic happens. Studying is useful, but it's only a small part of the overall process. |
|
|
I find it deeply reassuring that we have both independently ended up in more or less the same spot with language learning. I am just hoping this is not just very comfortable local minima!
I wonder what all this has to say about language schools, who seem even more stuck in the tarpit than Chomsky.
Edited by patrickwilken on 01 February 2014 at 2:05pm
1 person has voted this message useful
|
emk Diglot Moderator United States Joined 5541 days ago 2615 posts - 8806 votes Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map
| Message 936 of 1317 01 February 2014 at 8:38pm | IP Logged |
patrickwilken wrote:
That's great. It did occur to me that the dubbing might actually account for some of the perceived sexism too - depending on how the voice-overs work. |
|
|
No, the characters were always well-written.
Jeffers wrote:
Amazon.fr has a 7 disc edition for about 16 euros, and is supposed to have 26 episodes. The amazon.fr version only has French and Japanese (and French subtitles).
Does either of these match what you got? Do the French subtitles come close to the French audio? |
|
|
I have the Amazon.fr version. It has Japanese and French soundtracks, and two French subtitle tracks. The French subtitles correspond very roughly to the dialog: About half the lines are accurate, and the other half are a rephrasing of the same idea. It's not like Engrenages, where the French subs are pretty useless, or Buffy, where the French audio was extensively rewritten to maintain lip-sync and overall style, but where the subs were translated directly from English. The French audio for NGE is moderately challenging.
Current status: Page 116 of Tara Duncan just reading it here and there during the week.
2 persons have voted this message useful
|
You cannot post new topics in this forum - You cannot reply to topics in this forum - You cannot delete your posts in this forum You cannot edit your posts in this forum - You cannot create polls in this forum - You cannot vote in polls in this forum
This page was generated in 0.5000 seconds.
DHTML Menu By Milonic JavaScript
|