According to Keith Teare, founder and CEO of RealNames Corporation, "Has anybody considered what this means for Microsoft's partners and the future of technology? [...] If this decision [the MS split-up] were to discourage them from embracing early stage technologies it could seriously slow down the adoption of these technologies".
We can only guess about speech production. One educated guess does include a knowledge-base of (about 200 IIRC) different sounds -- I don't know if this "dictionary" of sounds is what he refer to as auditory memory.
which is in the same space as your short-term
We have not located any dedicated short-term memory (it's a term to cover the phenomenon that we seem to only remember 7 ± 2 "chunks" at once).
Some models picture the active neurons in the brain as constituting the conscious, which could be compared with (a superset of) the short-term memory, and thus, exclude the theory that it should be in a central place.
and working memory
Working memory is a model, and does not map to a location in the brain.
In short, I don't think this guy refers to science, but merely his own experience.
As said then the working memory is a model, and it includes a phonological loop which can store approximately 2 seconds of audio. This is believed to be the auditory short term memory.
Empirical evidence seems to show that there are many resource conflicts between this phonological loop and speech production, but more importantly for this guy (cause why would you need an auditory short-term memory while using your computer? at least on a properly designed system this should not be required!) the resource conflicts also seem to involve reading.
So for example if you had to say "scroll down" while reading a text in a browser, then that would probably make you stop reading, while saying these two words. But, with time this may become an automated process (associative learning), and each time your eyes reach the lower half of the browser-window, you'll say "scroll down" without even thinking about it, similar to how you currently probably press arrow-down or tweak the mouse-wheel w/o any conscious, even though the first 20 or so times you had to scroll, it'd disrupt your work-flow -- personally I have no knowledge if an "automated" voice action would give rise to resource conflicts with a conscious reading-process. Though I'm certain that I've experienced people e.g. saying "auch" or similar while reading a newspaper, without really being interrupted. A thing to keep in mind here is also that male and female brain wiring with regard to speaking is not identical. For example women are much more capable of speaking or reading while listening, which is near impossible for men.
But when all this is said, then let me also say that I think the author of this piece is completely wrong in his conclusion that speech recognition won't be of use.
It might not be useful for many of these one-key-stroke tasks, but it's definitely valuable for more complex tasks (like saying "Show the mail I just got", "Start the project I worked on yesterday", "Start a wordprocessor with an empty document" etc. -- all said while not using the application in question).
It may also serve as an extra input device when your hands are busy (e.g. in a game, where you want to change the perspective, weapon or zoom).
Actually, all tasks where the command you would have to give could be given unambiguously simply by thinking out-load, would be perfect candidates for speech recognition!
Tab'ing using a stack (was: They'll never get me)
on
Penguin2Apple
·
· Score: 1
Consider this: OS X comes with an alt-tab action, but it cycles through all windows in a circular list, rather than using a stack like Windows or most X11 window managers.
Probably you'd want to download Keyboard Maestro -- this gives you tab'ing using a stack rather than a cyclic list.
You can't really expect every little bit of the default installation to work exactly as you'd prefer, sometimes you have to tweak the settings yourself or download utilities that add the missing functionality -- I'd assume the same is the case for unix...
Picking a process to run is O(p) as long as no real-time processes are runnable (which is the vast majority of the time)
ehm... most of the time there are no processes to run, they all wait for a message/interrupt. When one arrives, they are moved to the list (priority queue) of running/ready processes.
The schedular always just pick the next in this list of running/ready processes, which is O(1). At no time does it need to look at the iddle processes, to pick one to run, as they don't need to be run... this is not Microsoft's cooperative multitasking...
For the Linux scheduler, switching processes is O(p), where p is the number of currently running processes. This new scheduler switches processes in O(1) time.
I haven't really worked with Linux, but despite my technical aversions toward this OS then I sure don't hope that switching processes is linear in time.
An OS maintain a list of running tasks (processes) and each time a task has used up its timeslice (which is around one tenth of a second) or volunterely given up the processor (because it has no more work to compute), then the next task is taken from the list (and the previously running task is appended to the end, unless it volunterely gave up the processor) -- this is "task switching" and the principle is often referred to as roundrobin. Something that most should be able to implement as a constant-time operation.
Often one would use a priority queue instead of a list, that is, the task each have a priority, and instead of adding tasks to the end of the list, they are inserted according to their priority (this can still be done in constant time by e.g. having an array of lists, where the index in the array is the priority).
Using priorities you ensure that tasks with high priority always get the processor when they need it (e.g. a mouse driver) and tasks with low priorities to never interfer with the users work (like a batch job running in the background).
The above is often called static scheduling, because the priorities are set once, and never touched. But this is often not desireable, because it's a tough job adjusting all those priorities, and perhaps not even possible on a multi user system.
So what do we do? Well... this is where black magic comes into play. One simple solution could be to boost the priority of all the tasks which doesn't use their full time slice (so they'll have a higher priority next time they run, because we assume that they won't run for long).
Many heuristics exist to try and make dynamic scheduling (priority rearranging) fair. Often these heuristics only rearrange priorities a few times pr. minute, based on statistics on how the tasks have behaved the last n seconds.
Since this process involve all the running tasks, then it is in nature linear in time (but of course some tricks exist). But as said, this algorithm runs only a few times pr. minute, but the result affect the entire fairness/responsiveness of the machine. So time is probably better spent crafting a better heuristic than making it O(1) instead of O(n).
Here are two articles that relate to the work done with NMDA receptors (from late '96). IMHO they are rather convincing of the role that synaptic strengthening plays in the process of learning.
The first article also tells that they were able to translate the activation pattern in hippocampus to the spatial location of the mouse (while it was swimming in Morris water maze).
It just goes to show how little real world experience students have
Apparently the author hasn't been a student himself. If so, he'd at least have learned not to make general accusations based on a single article on the internet!
Re:Its the squeeky wheel that gets the most attent
on
Interview With Linus
·
· Score: 1
Linux may be better than Windows at some points, but I wouldn't refer to it as technically superiour.
But this does not change the fact that the public should know about Microsoft the company:
how they deliberately break open standards (to make them semi-proprietary and/or useless and at times then introduce their own alternative closed Win-only alternative),
how they steal ideas, patents and even source code,
how they often play a big role in the limited succes of their competitors (OS/2, JAVA, DR-DOS etc.),
how they trade off comfort to get more sales (I no longer view it as incompetence that older versions of their programs can't understand files produced by the latest version),
how they neglect security issues,
add your own aversion here...
But I wouldn't expect this to come from an interview with Linus, this is not really his domain, and he would be too biased anyway...
Slashdot Mirror
Oh the irony...
We can only guess about speech production. One educated guess does include a knowledge-base of (about 200 IIRC) different sounds -- I don't know if this "dictionary" of sounds is what he refer to as auditory memory.
We have not located any dedicated short-term memory (it's a term to cover the phenomenon that we seem to only remember 7 ± 2 "chunks" at once).
Some models picture the active neurons in the brain as constituting the conscious, which could be compared with (a superset of) the short-term memory, and thus, exclude the theory that it should be in a central place.
Working memory is a model, and does not map to a location in the brain.In short, I don't think this guy refers to science, but merely his own experience.
As said then the working memory is a model, and it includes a phonological loop which can store approximately 2 seconds of audio. This is believed to be the auditory short term memory.
Empirical evidence seems to show that there are many resource conflicts between this phonological loop and speech production, but more importantly for this guy (cause why would you need an auditory short-term memory while using your computer? at least on a properly designed system this should not be required!) the resource conflicts also seem to involve reading.
So for example if you had to say "scroll down" while reading a text in a browser, then that would probably make you stop reading, while saying these two words. But, with time this may become an automated process (associative learning), and each time your eyes reach the lower half of the browser-window, you'll say "scroll down" without even thinking about it, similar to how you currently probably press arrow-down or tweak the mouse-wheel w/o any conscious, even though the first 20 or so times you had to scroll, it'd disrupt your work-flow -- personally I have no knowledge if an "automated" voice action would give rise to resource conflicts with a conscious reading-process. Though I'm certain that I've experienced people e.g. saying "auch" or similar while reading a newspaper, without really being interrupted. A thing to keep in mind here is also that male and female brain wiring with regard to speaking is not identical. For example women are much more capable of speaking or reading while listening, which is near impossible for men.
But when all this is said, then let me also say that I think the author of this piece is completely wrong in his conclusion that speech recognition won't be of use.
It might not be useful for many of these one-key-stroke tasks, but it's definitely valuable for more complex tasks (like saying "Show the mail I just got", "Start the project I worked on yesterday", "Start a wordprocessor with an empty document" etc. -- all said while not using the application in question).
It may also serve as an extra input device when your hands are busy (e.g. in a game, where you want to change the perspective, weapon or zoom).
Actually, all tasks where the command you would have to give could be given unambiguously simply by thinking out-load, would be perfect candidates for speech recognition!
Now we just need to show that N = 1... :-)
Probably you'd want to download Keyboard Maestro -- this gives you tab'ing using a stack rather than a cyclic list.
You can't really expect every little bit of the default installation to work exactly as you'd prefer, sometimes you have to tweak the settings yourself or download utilities that add the missing functionality -- I'd assume the same is the case for unix...
ehm... most of the time there are no processes to run, they all wait for a message/interrupt. When one arrives, they are moved to the list (priority queue) of running/ready processes.
The schedular always just pick the next in this list of running/ready processes, which is O(1). At no time does it need to look at the iddle processes, to pick one to run, as they don't need to be run... this is not Microsoft's cooperative multitasking...
hmm... the tuner in the settopbox in front of me is actually a lot smaller than the x86 (incl. cooling) in the pc which is also in front of me...
I haven't really worked with Linux, but despite my technical aversions toward this OS then I sure don't hope that switching processes is linear in time.
An OS maintain a list of running tasks (processes) and each time a task has used up its timeslice (which is around one tenth of a second) or volunterely given up the processor (because it has no more work to compute), then the next task is taken from the list (and the previously running task is appended to the end, unless it volunterely gave up the processor) -- this is "task switching" and the principle is often referred to as roundrobin. Something that most should be able to implement as a constant-time operation.
Often one would use a priority queue instead of a list, that is, the task each have a priority, and instead of adding tasks to the end of the list, they are inserted according to their priority (this can still be done in constant time by e.g. having an array of lists, where the index in the array is the priority).
Using priorities you ensure that tasks with high priority always get the processor when they need it (e.g. a mouse driver) and tasks with low priorities to never interfer with the users work (like a batch job running in the background).
The above is often called static scheduling, because the priorities are set once, and never touched. But this is often not desireable, because it's a tough job adjusting all those priorities, and perhaps not even possible on a multi user system.
So what do we do? Well... this is where black magic comes into play. One simple solution could be to boost the priority of all the tasks which doesn't use their full time slice (so they'll have a higher priority next time they run, because we assume that they won't run for long).
Many heuristics exist to try and make dynamic scheduling (priority rearranging) fair. Often these heuristics only rearrange priorities a few times pr. minute, based on statistics on how the tasks have behaved the last n seconds.
Since this process involve all the running tasks, then it is in nature linear in time (but of course some tricks exist). But as said, this algorithm runs only a few times pr. minute, but the result affect the entire fairness/responsiveness of the machine. So time is probably better spent crafting a better heuristic than making it O(1) instead of O(n).
Here are two articles that relate to the work done with NMDA receptors (from late '96). IMHO they are rather convincing of the role that synaptic strengthening plays in the process of learning.
The first article also tells that they were able to translate the activation pattern in hippocampus to the spatial location of the mouse (while it was swimming in Morris water maze).
Unfortunately those people are not aware that many of us also use the spoken language.
How would I ask people to attend my 1Ath birthday...
Apparently the author hasn't been a student himself. If so, he'd at least have learned not to make general accusations based on a single article on the internet!
Linux may be better than Windows at some points, but I wouldn't refer to it as technically superiour.
But this does not change the fact that the public should know about Microsoft the company:
But I wouldn't expect this to come from an interview with Linus, this is not really his domain, and he would be too biased anyway...