That's an example of a failure of defense in depth. Why shouldn't gethostbyname() be executed in the kernel? Seems like a reasonable place to execute system calls no? The reason is, you want to execute everything with the lowest possible privileges. If there is an exploitable bug, the exploit doesn't get you very far.
The world is a ball of rock with a thin veneer of feces. They're everywhere. Your phone probably has less on it than anything else because you wipe the screen occasionally.
There seems to be some revisionist history going on. When Apple released the first iPad I remember talk about it replacing netbooks (Slashdot was very skeptical) but not notebooks and certainly not phones. The various tablets seem to have killed netbooks very dead and left phones and notebooks mostly alone. I'm sure Apple and Samsung, who both make phones, tablets and notebooks, are very happy with that outcome.
That was my question. Where did the cops get a bomb? Do American cops stock grenades now? Or did some clever officer mix together some common household products he found around the station?
"A.I. that decide the rules of the game and discover winning conditions; those are a good 60 ~ 100 years away."
Nope. There's at least one example of a company that developed an AI that can be given any random game and it will figure out the rules and how to win. They let it play Nintendo. Google bought them. The more general field is called reinforcement learning.
Except they didn't? Theranos operated fraudulently for some time, requiring a lot of tests to be redone. They also seem to have lied to investors, customers and the public about what they were and could do, creating a serious distortion of a free market, where accurate information is available to all.
Is it? Considering all the hand wringing over the ability of companies to attract "talent", you'd think that retaining what you've got would be worth a bit of non-sociopathic behaviour.
If I make an unsafe left turn across oncoming traffic on a highway the cops are not going to accept the excuse that I didn't want to wait for a bigger break in traffic. If the intersection is regularly so busy that it's not possible for vehicles to make a safe turn then it needs an overpass or lights. I don't know where this accident occurred, but I doubt there's a line in the highway code that says you have to make safe left turns unless you're a transport truck.
Shit happens, but my point is that this accident wasn't "Tesla's autopilot screwed up" as is being reported. It's "(1) transport truck made an illegal and dangerous turn, (2) Tesla driver was watching Harry Potter instead of driving, (3) Tesla autopilot failed to prevent accident caused by these two chuckleheads."
That seems to be an overlooked bit of this case. From all the accounts I've read, the truck driver was at fault for making a left turn into oncoming traffic. The truck driver made an illegal turn, then both the Tesla autopilot and the driver failed to avoid the danger. Blaming that on the autopilot seems a pretty big stretch.
Slashdot in particular seems to have this weird cognitive dissonance around AI in general, and automatic car features in particular: the insistence that a computer cannot do complex tasks as well as or better than a human, and the judgement that such a system has failed unless it performs absolutely perfectly at all times.
Not advanced at all. I just taught a workshop on neural networks using identifying both the location and value of several handwritten digits scattered randomly in a larger image as an example problem.
Fonts are easier because they're much more consistent than handwritten characters.
It shouldn't. It's pretty trivial. You could do it yourself by taking a picture then using that eyedropper tool on it.
You might not get the results you expect though. The color the camera sees will depend a lot on the illumination color. And your eyes don't faithfully report the actual color of things.
I was trained to sail solo. Part of that training is the construction of an autopilot. It consists of a length of rope and the ability to tie a few knots.
Wearing a helmet comes with a significant penalty in the form of inconvenience. Putting cameras on your car does too, but much less so for a reasonable fraction of the populace than wearing a helmet.
"When aircraft stalls it just falls from the sky. "
No it doesn't. Any competently designed plane will pitch down when it stalls, pick up speed, and not be stalled anymore. The Air France flight wasn't flying along at altitude, stalled, then plunged into the ocean. Due to instrument failure and pilot error it took a long slow path down; long and slow enough that neither pilot realized what was happening until it was too late.
Not sure what you're doing, but you're doing it wrong. The DICOM standard includes very specific tags for identifying orientation unambiguously. In hundreds of thousands of images over a decade and a half I've never seen a DICOM file from an image acquisition system that didn't properly implement them.
There have been a few other papers criticising FWE clustering lately. It's always struck me as kind of an iffy concept. Even the simpler non-clustering techniques, although they seem to do more or less what they advertise, really should be regarded as exploratory and checked by proper hypothesis driven replication studies.
If that's true then it suggests perhaps we should be careful how much we do to "encourage" girls. A common tactic to get more women into a field is to create special programs, scholarships, quotas, that reduce the need for the females to compete in the general pool. That might have unintended consequences.
Maybe you should look around more. Lots of wealthy people retire from their jobs to do things like hobby farming or frequenting wilderness retreats. There's a guy who runs a pizza parlour down the street who retired from some kind of executive position. He scrubs the toilets there, and also the kitchen, and waits tables. I know another guy who retired from aerospace engineering to teach sailing. I've seen him scrub a marine head. And a deck.
That's an example of a failure of defense in depth. Why shouldn't gethostbyname() be executed in the kernel? Seems like a reasonable place to execute system calls no? The reason is, you want to execute everything with the lowest possible privileges. If there is an exploitable bug, the exploit doesn't get you very far.
The world is a ball of rock with a thin veneer of feces. They're everywhere. Your phone probably has less on it than anything else because you wipe the screen occasionally.
There seems to be some revisionist history going on. When Apple released the first iPad I remember talk about it replacing netbooks (Slashdot was very skeptical) but not notebooks and certainly not phones. The various tablets seem to have killed netbooks very dead and left phones and notebooks mostly alone. I'm sure Apple and Samsung, who both make phones, tablets and notebooks, are very happy with that outcome.
That was my question. Where did the cops get a bomb? Do American cops stock grenades now? Or did some clever officer mix together some common household products he found around the station?
"A.I. that decide the rules of the game and discover winning conditions; those are a good 60 ~ 100 years away."
Nope. There's at least one example of a company that developed an AI that can be given any random game and it will figure out the rules and how to win. They let it play Nintendo. Google bought them. The more general field is called reinforcement learning.
Google does the same thing.
You know John Oliver televised "Drumpf" because Trump repeatedly mocked John Stewart for changing his name, right?
Except they didn't? Theranos operated fraudulently for some time, requiring a lot of tests to be redone. They also seem to have lied to investors, customers and the public about what they were and could do, creating a serious distortion of a free market, where accurate information is available to all.
Is it? Considering all the hand wringing over the ability of companies to attract "talent", you'd think that retaining what you've got would be worth a bit of non-sociopathic behaviour.
I'm bored. Can I have some of what you're smoking?
In English autopilot doesn't necessarily mean what it's etymological roots do in Greek.
Even in isolation, the older uses of pilot are more along the lines of "helmsman." Helmsmen steer the ship. Someone else tells them where to steer it.
If I make an unsafe left turn across oncoming traffic on a highway the cops are not going to accept the excuse that I didn't want to wait for a bigger break in traffic. If the intersection is regularly so busy that it's not possible for vehicles to make a safe turn then it needs an overpass or lights. I don't know where this accident occurred, but I doubt there's a line in the highway code that says you have to make safe left turns unless you're a transport truck.
Shit happens, but my point is that this accident wasn't "Tesla's autopilot screwed up" as is being reported. It's "(1) transport truck made an illegal and dangerous turn, (2) Tesla driver was watching Harry Potter instead of driving, (3) Tesla autopilot failed to prevent accident caused by these two chuckleheads."
"The truck probably didn't see the car either."
That seems to be an overlooked bit of this case. From all the accounts I've read, the truck driver was at fault for making a left turn into oncoming traffic. The truck driver made an illegal turn, then both the Tesla autopilot and the driver failed to avoid the danger. Blaming that on the autopilot seems a pretty big stretch.
Slashdot in particular seems to have this weird cognitive dissonance around AI in general, and automatic car features in particular: the insistence that a computer cannot do complex tasks as well as or better than a human, and the judgement that such a system has failed unless it performs absolutely perfectly at all times.
We have a hard time communicating because anonymous cowards keep making up definitions for words.
Autopilot has never meant start it and forget it, and nothing called an autopilot has ever had that capability.
If Tesla called it "self-drive mode" then you might have a point. They don't.
Not advanced at all. I just taught a workshop on neural networks using identifying both the location and value of several handwritten digits scattered randomly in a larger image as an example problem.
Fonts are easier because they're much more consistent than handwritten characters.
It shouldn't. It's pretty trivial. You could do it yourself by taking a picture then using that eyedropper tool on it.
You might not get the results you expect though. The color the camera sees will depend a lot on the illumination color. And your eyes don't faithfully report the actual color of things.
Elon may also be betting on technology getting good enough to no longer need LIDAR. Such as this: http://www.ti.com/lsds/ti/sens....
I was trained to sail solo. Part of that training is the construction of an autopilot. It consists of a length of rope and the ability to tie a few knots.
Wearing a helmet comes with a significant penalty in the form of inconvenience. Putting cameras on your car does too, but much less so for a reasonable fraction of the populace than wearing a helmet.
"When aircraft stalls it just falls from the sky. "
No it doesn't. Any competently designed plane will pitch down when it stalls, pick up speed, and not be stalled anymore. The Air France flight wasn't flying along at altitude, stalled, then plunged into the ocean. Due to instrument failure and pilot error it took a long slow path down; long and slow enough that neither pilot realized what was happening until it was too late.
Not sure what you're doing, but you're doing it wrong. The DICOM standard includes very specific tags for identifying orientation unambiguously. In hundreds of thousands of images over a decade and a half I've never seen a DICOM file from an image acquisition system that didn't properly implement them.
http://dicom.nema.org/medical/...
Also, if you can't figure out all the directions except L/R with the skull stripped, you should probably take an anatomy class. Or look at a scan.
There have been a few other papers criticising FWE clustering lately. It's always struck me as kind of an iffy concept. Even the simpler non-clustering techniques, although they seem to do more or less what they advertise, really should be regarded as exploratory and checked by proper hypothesis driven replication studies.
But he did hire the creepy sounding woman? They disguised voices both ways.
If that's true then it suggests perhaps we should be careful how much we do to "encourage" girls. A common tactic to get more women into a field is to create special programs, scholarships, quotas, that reduce the need for the females to compete in the general pool. That might have unintended consequences.
Maybe you should look around more. Lots of wealthy people retire from their jobs to do things like hobby farming or frequenting wilderness retreats. There's a guy who runs a pizza parlour down the street who retired from some kind of executive position. He scrubs the toilets there, and also the kitchen, and waits tables. I know another guy who retired from aerospace engineering to teach sailing. I've seen him scrub a marine head. And a deck.