Strong AI hasn't really progressed since it was introduced (they're still arguing over what intelligence is, much less how to create it!), but weak AI has made some pretty good strides. For instance, I work on software that can read medical images and render a diagnosis in lieu of a second radiologist (this is called computer-assisted diagnosis). 15 years ago, this would not have been possible.
Not that this would be a good idea. Greed is a desire to further one's own position. We owe much of our lifestyle to it; capitalism has transformed it into a powerful force for progress. Remove it from society and society will stagnate. It's only a bad thing if it starts trampling on others' rights.
I'm similarly against removing any "undesirable" traits. Who gets to decide what is "desirable"? The traits people view as "proper" tend to be social norms, and do change over time. What gives that person the right to impose his own view of morality on everyone, before they even acquire the capacity to choose their own morality for themselves? What would happen to our individuality?
If you can't move forward, you'll surely fall behind. We have new ways of doing things which have the potential to make all of our lives easier. Our tools must evolve with us, or we must find new ones.
That's how they initially price it. Then another manufacturer comes along and charges X - 10%, then the first lowers their price, then they eventually approach Y, the cost of making the machine. Then someone else discovers how to do it cheaper and lowers the price further than competitors can to gain an advantage, and the price drops some more...
At least, that's how it's supposed to work. Works pretty well in tech., not sure if it's the same in medicine.
The GP is correct... functional MRI measures blood oxygenation level dependent (BOLD) response - that is, the change in paramagnetism induced by oxygenated hemoglobin. Active neurons require more oxygen than inactive neurons, so oxygenated blood is delivered to them more rapidly (the hemodynamic response). This induces a local shift in magnetic permeability (from paramagnetic to diamagnetic) which can be picked up by the scanner.
Whether the BOLD signal truly correlates well with neural activity is still a matter of contention within the medical community.
Many authors who feel as you do go ahead and post their papers on their websites for all to view anyway once they're published. Google Scholar now indexes them too. I agree completely on the journals, but in the meantime, you can always try finding papers this way.
In other words, Google is yet another organization jumping on the tensor mining bandwagon prior to assessing its merits and pitfalls? If they're using the same algorithms I think they're using, Google is going to have a heck of a time with the efficiency, considering the scale of their dataset. The 502 error I get when I attempt to access it isn't encouraging either.
This is Microsoft we're talking about. When they can't find a country that has exactly the laws they want, they'll make one (by copying the USA prior to making their "customizations"). The documented laws will be easy to understand (but undocumented laws will be used to thoroughly shaft the populace), they will have the friendliest secret police force in the world (to ensure the "best possible user disappearance experience" for dissenters), everyone will use their Live accounts to conduct any business within the country, and the entire economy will collapse about once a week.
Think of it as more beautiful music and an opportunity to grow as a pianist rather than as a chore. The FF pieces have always been among my favorite not only to hear, but to play as well.
Of the factors listed, time of day is the most important to me. I generally have a horrible time maintaining my concentration around 3-4 PM (actually, I have a hard time even staying awake in the office around that time), but I can code like a ninja at all other times of the day, particularly in the morning. I hit my peak concentration at about 11 AM. That goes not just for coding, but for taking exams, giving presentations, playing at concerts, etc.
(There is irony that I'm posting this on Slashdot @ 11:20 AM)
Except for the increased effort that many people are unwilling to make.
If you learn easily, you tend to view learning as an interesting exploration rather than a chore. If you don't learn easily, chances are you'll find something else to value rather than do something that does not come easily to you.
We're classifying by subject ("whose brain is this based on the other scans in the dataset?"), not by terrorist thoughts. The problem you're discussing is much harder.
Maybe this is different in EEG, but in the context of fMRI, subject is likely the easiest variable to classify on. I've done some experiments on it myself as part of my doctoral work. Even something as simple as wavelet processing the data and performing k-nearest neighbor classification yielded 96% accuracy on a motor task fMRI dataset of moderate size, and that was just a baseline method I was using to compare a better one to! In fact, the effect of the task was being obscured by the subject, and we had to remove the subject means from the data before we could attempt to classify on task (what we really wanted to do) with any reasonable degree of success.
It's probably the nature of fMRI image acquisition itself that has prevented this from being used as a biometric, for obvious reasons (asking people to submit to fMRI scans at airports is neither reasonable nor practical, and even EEG requires too much preparation to use as a general screen, considering how long lines get already). It's definitely not the difficulty of classification holding it back.
If I can find a corpus of geographically labeled text documents, I'll run a few text mining algorithms on the letter and see what pops up (yes, your writing can now give away things that you never thought possible, at least probabilistically).
Apparently the author likely has an ESTJ personality in the Myers-Briggs system and is probably male.
Slashdot Mirror
Strong AI hasn't really progressed since it was introduced (they're still arguing over what intelligence is, much less how to create it!), but weak AI has made some pretty good strides. For instance, I work on software that can read medical images and render a diagnosis in lieu of a second radiologist (this is called computer-assisted diagnosis). 15 years ago, this would not have been possible.
Not that this would be a good idea. Greed is a desire to further one's own position. We owe much of our lifestyle to it; capitalism has transformed it into a powerful force for progress. Remove it from society and society will stagnate. It's only a bad thing if it starts trampling on others' rights.
I'm similarly against removing any "undesirable" traits. Who gets to decide what is "desirable"? The traits people view as "proper" tend to be social norms, and do change over time. What gives that person the right to impose his own view of morality on everyone, before they even acquire the capacity to choose their own morality for themselves? What would happen to our individuality?
Would be kind of nice if hash_map could finally make it into std, though. Maybe even with a consistent include file across different platforms :)
If you can't move forward, you'll surely fall behind. We have new ways of doing things which have the potential to make all of our lives easier. Our tools must evolve with us, or we must find new ones.
That's how they initially price it. Then another manufacturer comes along and charges X - 10%, then the first lowers their price, then they eventually approach Y, the cost of making the machine. Then someone else discovers how to do it cheaper and lowers the price further than competitors can to gain an advantage, and the price drops some more...
At least, that's how it's supposed to work. Works pretty well in tech., not sure if it's the same in medicine.
Neither writing a song nor writing a book are trivial endeavors. Actually, some scientific research takes less time and effort than either to publish.
Which makes one wonder what is taking them so long to finalize it.
You're correct. Tenacity matters much more than intelligence in a doctoral program.
Bubble sort is patented?
Yes, but the people in charge still haven't listened!
The GP is correct... functional MRI measures blood oxygenation level dependent (BOLD) response - that is, the change in paramagnetism induced by oxygenated hemoglobin. Active neurons require more oxygen than inactive neurons, so oxygenated blood is delivered to them more rapidly (the hemodynamic response). This induces a local shift in magnetic permeability (from paramagnetic to diamagnetic) which can be picked up by the scanner.
Whether the BOLD signal truly correlates well with neural activity is still a matter of contention within the medical community.
...And a working knowledge of how high-level cognitive phenomena emerge from the firings of these neurons...
What about the laws of thermodynamics? They seem to establish a default state of the universe which may only be altered with an input of energy.
The size of the market for commercial spaceflight is limited by the lack of destinations.
Many authors who feel as you do go ahead and post their papers on their websites for all to view anyway once they're published. Google Scholar now indexes them too. I agree completely on the journals, but in the meantime, you can always try finding papers this way.
In other words, Google is yet another organization jumping on the tensor mining bandwagon prior to assessing its merits and pitfalls? If they're using the same algorithms I think they're using, Google is going to have a heck of a time with the efficiency, considering the scale of their dataset. The 502 error I get when I attempt to access it isn't encouraging either.
This is Microsoft we're talking about. When they can't find a country that has exactly the laws they want, they'll make one (by copying the USA prior to making their "customizations"). The documented laws will be easy to understand (but undocumented laws will be used to thoroughly shaft the populace), they will have the friendliest secret police force in the world (to ensure the "best possible user disappearance experience" for dissenters), everyone will use their Live accounts to conduct any business within the country, and the entire economy will collapse about once a week.
Think of it as more beautiful music and an opportunity to grow as a pianist rather than as a chore. The FF pieces have always been among my favorite not only to hear, but to play as well.
Of the factors listed, time of day is the most important to me. I generally have a horrible time maintaining my concentration around 3-4 PM (actually, I have a hard time even staying awake in the office around that time), but I can code like a ninja at all other times of the day, particularly in the morning. I hit my peak concentration at about 11 AM. That goes not just for coding, but for taking exams, giving presentations, playing at concerts, etc.
(There is irony that I'm posting this on Slashdot @ 11:20 AM)
Except for the increased effort that many people are unwilling to make.
If you learn easily, you tend to view learning as an interesting exploration rather than a chore. If you don't learn easily, chances are you'll find something else to value rather than do something that does not come easily to you.
We're classifying by subject ("whose brain is this based on the other scans in the dataset?"), not by terrorist thoughts. The problem you're discussing is much harder.
The (mainstream) Internet Era has been around for some 15 years now and it hasn't yet.
Maybe this is different in EEG, but in the context of fMRI, subject is likely the easiest variable to classify on. I've done some experiments on it myself as part of my doctoral work. Even something as simple as wavelet processing the data and performing k-nearest neighbor classification yielded 96% accuracy on a motor task fMRI dataset of moderate size, and that was just a baseline method I was using to compare a better one to! In fact, the effect of the task was being obscured by the subject, and we had to remove the subject means from the data before we could attempt to classify on task (what we really wanted to do) with any reasonable degree of success.
It's probably the nature of fMRI image acquisition itself that has prevented this from being used as a biometric, for obvious reasons (asking people to submit to fMRI scans at airports is neither reasonable nor practical, and even EEG requires too much preparation to use as a general screen, considering how long lines get already). It's definitely not the difficulty of classification holding it back.
If I can find a corpus of geographically labeled text documents, I'll run a few text mining algorithms on the letter and see what pops up (yes, your writing can now give away things that you never thought possible, at least probabilistically).
Apparently the author likely has an ESTJ personality in the Myers-Briggs system and is probably male.
Where were you in January?