I don't think this has anything to do with the lack of reasoning or putting things in context and much more with a statistical glitch.
The state of the art in object detection is around 50% mAP, which is not that great. Even on untaylored images, you have many some false alarms and misdetections, so it no surprise that by modifying images in a way that separates them completely from the training data, it leads to some strange false alarms.
I think the authors could just have looked at the validation set and exhibit the images that are performing the worst and draw the same conclusion that computer vision sucks, if that's what they really wanted to do.
The fact is, although computer vision systems are nowhere near perfect, they are improving are an impressive rate. Even without ideological statements about the necessity of reasoning or what not.
There is no shortage of talented people, and as such, there is no need to make room for everyone. There is a fair number of excellent developers with a thick skin, and it's more than enough to make a project like the linux kernel progress at a steady pace. There is no need to broaden the circle to people that are supposedly good but don't like to have their feelings hurt.
The kernel documentation is atrocious and somebody wasted time to produce this instead of improving the project.
I've always though linux was a technical project by very talented people, not a pile of bullshit superficial texts by some teens in need of recognition.
It takes only one guy with the right capabilities and stupid enough to do it. History has proven that there are plenty of such people. You can be sure that there are plenty of high level military officers in many countries that are day dreaming of something from Screamers, and will do anything that is in their power to make it a reality...
I know your domestic politics is none of our concern, but seen from the rest of the world, a president that tweets he can pardon himself looks pretty damn ridiculous and is not really a good negotiation signal...
A cynic might say the world needs a massive recession or war -- carbon emissions actually went down in the 2008-9 recession.
The world would undergo a massive recession and the accompanying war when the decline in energy production due to the end of fossil fuel will violently meet the growth in energy demand due to a growing population with higher standard of living. The exact date is up to prediction, though.
Now think of all similar algorithms that will make decisions and the data they have been trained on. For example, the algorithm that process admission to a university, or the algorithm that computes the cost of your health insurance. You want these to have been trained on data that are favourable to you or at least neutral, but you'll never now unless the training data are public. Actually, and contrarily to other algorithms, with machine learning, you don't really care about the algorithm (with its millions parameters, it's completely opaque), you care about the training data.
Isn't that a question of density of population? Like in population has grown so much that everything is getting scarcer for everybody. I know the US are huge (I'm European), but nonetheless, some areas are so overcrowded that it's impossible but for the wealthiest to buy a property. The same goes for jobs: there are already a million people with the same skills than you but better at them, and available from all over the world thanks to globalization. It's becoming more and more difficult to stand out and not just be useless. You can make similar reasoning for almost all the things that people in their 30s have more difficulties to obtain than their parent. Isn't that all linked to the size of the population compared to the size of our little planet?
Canonical (or other companies) should offer a service that does code reviews and certifies that a specific revision is malware free for a small amount of money.
Sure, if you're developing a free software, you probably do not have the money to do so, but you could always ask the community to fund the certification.
Or Canonical could set up a voting system where the most voted apps get certified periodically.
I do my OOP exams on paper because I don't care if you can or cannot write correctly in $LANGUAGE. The IDE and stackoverflow will eventually correct your code syntax. What I want to test is whether you can understand what a piece of complex undocumented code does, how do the OOP concepts work in practice* and whether you are capable of designing a software architecture that uses said concepts to solve a non-trivial problem.
If you think a good programmer is somebody that can translate a pseudo-code algorithm into his/her favourite language blindfolded, think again.There will be an AI available on the market that does that quicker and better than you before you hit 40.
* Here's an example in Java: class A {
public int x;
public A() {
this.x = 0;
inc();
}
public void inc() {
this.x += 1;
} }
class B extends A {
public B() {
super();
inc();
}
public void inc() {
this.x += 5;
}
public static void main(String[] args) {
B b = new B();
System.out.println(b.x);
} }
What does that code print and why? You'd be surprised how many students fail a question like that. Now, this is typically a paper question that doesn't make much sense if computers are allowed. However, I think it's a question that makes a clear distinction between the students that truly understood what polymorphism is and those who did not completely, which is a pretty good indication of who is going to design correct programs and who is not....
No it's not! Many jobs are about the pleasure of producing something or contributing to society and not about suffering for the pleasure of a soulless dude that has your survival at the disposal of his money!
Well, France has a lot of overseas territories which might or might not contain oil. There are even known deposits in the Mozambique channel where France have very large territorial waters.
In fact, the move is more signalling that France will not exploit those resources in the future than stopping current exploitation.
It's the same for music, painting, sculpture, film, etc. The art creation economy has always been that way, because it comes from an intrinsic property: When you need 100 000 people to give you money in exchange of something in order to survive, there is absolutely zero chance that more than a handful can live that way.
I had exactly the same mental image as you. In French, both the continent and the moon are spelled "Europe", which is as confusing. I had to read the summary twice to remember that Europa is the moon.
I suggest we add Europa as the newest 28th state of the EU (now that the brits are out).
Since it's a free launch, they could have proposed academics to send whatever experiments they want to put into mars orbit for free, with no guarantee of success. I'm pretty sure a lot of professors would have loved to have students come up with a micro satellite design and build it as part of a project. It's sad if it blows up on launch, but it's not critical, and if it works, it's cool for the students and maybe you get interesting measurements as a byproduct.
As usual, it's PR winning against anything else that would have been more useful...
Slashdot Mirror
I don't think this has anything to do with the lack of reasoning or putting things in context and much more with a statistical glitch.
The state of the art in object detection is around 50% mAP, which is not that great. Even on untaylored images, you have many some false alarms and misdetections, so it no surprise that by modifying images in a way that separates them completely from the training data, it leads to some strange false alarms.
I think the authors could just have looked at the validation set and exhibit the images that are performing the worst and draw the same conclusion that computer vision sucks, if that's what they really wanted to do.
The fact is, although computer vision systems are nowhere near perfect, they are improving are an impressive rate. Even without ideological statements about the necessity of reasoning or what not.
There is no shortage of talented people, and as such, there is no need to make room for everyone. There is a fair number of excellent developers with a thick skin, and it's more than enough to make a project like the linux kernel progress at a steady pace. There is no need to broaden the circle to people that are supposedly good but don't like to have their feelings hurt.
WTF is this?
The kernel documentation is atrocious and somebody wasted time to produce this instead of improving the project.
I've always though linux was a technical project by very talented people, not a pile of bullshit superficial texts by some teens in need of recognition.
Rejoice! You're going to pay more after leaving the EU and the introduction of a tariff on the imported french electricity.
Well, that gives you an idea of how this card generation got its performances boost.
More than 200 comments on the optimal size and design of plugs is why I love slashdot. This is truly stuff that matters.
It takes only one guy with the right capabilities and stupid enough to do it. History has proven that there are plenty of such people. You can be sure that there are plenty of high level military officers in many countries that are day dreaming of something from Screamers, and will do anything that is in their power to make it a reality...
I know your domestic politics is none of our concern, but seen from the rest of the world, a president that tweets he can pardon himself looks pretty damn ridiculous and is not really a good negotiation signal...
A cynic might say the world needs a massive recession or war -- carbon emissions actually went down in the 2008-9 recession.
The world would undergo a massive recession and the accompanying war when the decline in energy production due to the end of fossil fuel will violently meet the growth in energy demand due to a growing population with higher standard of living. The exact date is up to prediction, though.
In other news, Apple wants to be the only one to be able to track its demographic to perform targeted advertising.
That's the point!
Now think of all similar algorithms that will make decisions and the data they have been trained on. For example, the algorithm that process admission to a university, or the algorithm that computes the cost of your health insurance. You want these to have been trained on data that are favourable to you or at least neutral, but you'll never now unless the training data are public. Actually, and contrarily to other algorithms, with machine learning, you don't really care about the algorithm (with its millions parameters, it's completely opaque), you care about the training data.
Isn't that a question of density of population? Like in population has grown so much that everything is getting scarcer for everybody. I know the US are huge (I'm European), but nonetheless, some areas are so overcrowded that it's impossible but for the wealthiest to buy a property. The same goes for jobs: there are already a million people with the same skills than you but better at them, and available from all over the world thanks to globalization. It's becoming more and more difficult to stand out and not just be useless. You can make similar reasoning for almost all the things that people in their 30s have more difficulties to obtain than their parent. Isn't that all linked to the size of the population compared to the size of our little planet?
Peak load in Europe is over 500GW, so yeah, 18MW do not sound exactly game changing...
https://docstore.entsoe.eu/Doc...
Canonical (or other companies) should offer a service that does code reviews and certifies that a specific revision is malware free for a small amount of money.
Sure, if you're developing a free software, you probably do not have the money to do so, but you could always ask the community to fund the certification.
Or Canonical could set up a voting system where the most voted apps get certified periodically.
There are plenty of solutions to this problem.
I do my OOP exams on paper because I don't care if you can or cannot write correctly in $LANGUAGE. The IDE and stackoverflow will eventually correct your code syntax. What I want to test is whether you can understand what a piece of complex undocumented code does, how do the OOP concepts work in practice* and whether you are capable of designing a software architecture that uses said concepts to solve a non-trivial problem.
If you think a good programmer is somebody that can translate a pseudo-code algorithm into his/her favourite language blindfolded, think again.There will be an AI available on the market that does that quicker and better than you before you hit 40.
* Here's an example in Java:
class A {
public int x;
public A() {
this.x = 0;
inc();
}
public void inc() {
this.x += 1;
}
}
class B extends A {
public B() {
super();
inc();
}
public void inc() {
this.x += 5;
}
public static void main(String[] args) {
B b = new B();
System.out.println(b.x);
}
}
What does that code print and why? You'd be surprised how many students fail a question like that. Now, this is typically a paper question that doesn't make much sense if computers are allowed. However, I think it's a question that makes a clear distinction between the students that truly understood what polymorphism is and those who did not completely, which is a pretty good indication of who is going to design correct programs and who is not....
You should check "Neural Turing machine" and "Differentiable neural computer".
Isn't that any job?
No it's not! Many jobs are about the pleasure of producing something or contributing to society and not about suffering for the pleasure of a soulless dude that has your survival at the disposal of his money!
You put value in people suffering for your pleasure?
Take that, Transparent Aluminum!
Well, France has a lot of overseas territories which might or might not contain oil. There are even known deposits in the Mozambique channel where France have very large territorial waters.
In fact, the move is more signalling that France will not exploit those resources in the future than stopping current exploitation.
It's the same for music, painting, sculpture, film, etc. The art creation economy has always been that way, because it comes from an intrinsic property: When you need 100 000 people to give you money in exchange of something in order to survive, there is absolutely zero chance that more than a handful can live that way.
Why should it be different on the internet?
I had exactly the same mental image as you. In French, both the continent and the moon are spelled "Europe", which is as confusing. I had to read the summary twice to remember that Europa is the moon.
I suggest we add Europa as the newest 28th state of the EU (now that the brits are out).
I honestly read that as "they could have proposed to send academics".
Better start with lawyers, then :)
Since it's a free launch, they could have proposed academics to send whatever experiments they want to put into mars orbit for free, with no guarantee of success. I'm pretty sure a lot of professors would have loved to have students come up with a micro satellite design and build it as part of a project. It's sad if it blows up on launch, but it's not critical, and if it works, it's cool for the students and maybe you get interesting measurements as a byproduct.
As usual, it's PR winning against anything else that would have been more useful...
Thanks you so much, AC!