The original version was trained on the whole map, then the second version was trained to see only the camera view. In training, it managed to get 95% as good as the whole-map version.
The human had time to prepare a counter-strategy to DeepMind, that is really the big variable that changed.
what was the real tipping point was that DeepMind could see anywhere where a unit was located. Humans however can only see a "screen at a time". When DeepMind's makers went back and implemented "screen at a time" limitation, DeepMind was easily fooled again
That aspect was over-hyped. In the set that was won by the human (where the screen was limited), the human had time to analyze the playing style of the computer, and come up with a good counter-strategy. That is the main reason the computer lost.
Limiting the screen reduced the skill of the computer a few percentages, but it wasn't the only or (in my opinion) the main difference.
The difference between Jeopardy and the Starcraft AI is that the Starcraft AI wouldn't have come close to beating a human if it weren't for the inhuman precision.
The strategies the computer came up with were lousy (and very map-dependant), but it was able to compensate by having extremely precise movements.
We don't know exactly why Komincz won this game after losing the previous five
You could know if you'd watch the games. In the first set, DeepMind won with inhumanly superior micro. It was really cool, but computers have been better at micro for a long time. Speed and precision are things computers are good at, that's why we have aimbots.
In the second set, the human readjusted, and thought of strategies that would defend against the superior micro (by building more powerful units), while taking advantage of the computer's weaknesses (poor knowledge of army compositions, weak knowledge of positioning, and seemingly no object permanence: once enemy units are out of view, it has no idea where they are or if they exist).
Unless you believe there are only four continents "continent" being nonstandard. The Wikipedia page doesn't mention it, but some people consider Central America to be a separate continent, as well.
Shouldn't you be speaking French or something? True, in Spanish they call it futbol but they also say "pie" instead of foot, so I don't know what you're complaining about. Wait until you realize they call it "bano" instead of bathroom.
If Oracle can claim copyright on simply the function names of Java,
Because they've licensed it to the world, just like every other programming language worth using. Python APIs are just as copyright as Java's, but again, there is a license.
In the post-Oracle vs Google world, make sure you have a license for any APIs you are using.
The thing I've recently heard about Java is that you are subject to Oracle's random whims. Right now, you can get and use the runtime environment and development environment for free, but you don't know if they will randomly decide to charge you a ton of money to use it and send an army of lawyers after you.
Between it's different versions, the security problems this brings, etc., it's dying fast in the professional environment
What security problems come with Java?
I don't see a lot of JSP servers, either
JSP is legacy now, but a lot of companies are using Java on the backend in a web services model. The frontend can be in Angular, or React, or whatever. It's a good choice when you want stable, cheap developers.
As my own professional opinion, I would say that Java is better for writing backend APIs than Node/Express.
We do not have the capability to get people to the Moon
This is a ridiculous statement. We have the technology to go to the moon, technology has advanced since the '70s. It's just a matter of whether we want to do it or not.
they've tricked a bunch of people into thinking that the flu shot is a vaccine, and that having everybody take it would create "herd immunity." Even though it clearly doesn't; the success rate is far too low to create that effect.
It's not a matter of smart or stupid here. Even a very smart person will come to the wrong conclusion if they have bad data.
When people are ignorant of statistics, they are unable to collect good information, and unable to figure out what is true among multiple interpretations. The correlation is real, what do they decide?
I have given up the fight against the autocomplete function on my smartphone already. There is no way to turn it off.
You can install an alternative keyboard that doesn't have auto-complete. If you want more specific advice as a shortcut, on Android, install the Korean keyboard (and of course only use the English half). It will give you suggestions in the top bar, but won't automatically complete anything for you unless you gesture for it.
If you are very picky about what you want the keyboard to do, then you can always build your own.
The reason they used Python is because the most important use cases are already covered in Java or C++ with things like autocomplete. Adding AI doesn't give much improvement to what already exists. But since Python is dynamic, you can't know the type at compile time. That is why you need to use these extreme solutions for Python.
Slashdot Mirror
The original version was trained on the whole map, then the second version was trained to see only the camera view. In training, it managed to get 95% as good as the whole-map version.
The human had time to prepare a counter-strategy to DeepMind, that is really the big variable that changed.
what was the real tipping point was that DeepMind could see anywhere where a unit was located. Humans however can only see a "screen at a time". When DeepMind's makers went back and implemented "screen at a time" limitation, DeepMind was easily fooled again
That aspect was over-hyped. In the set that was won by the human (where the screen was limited), the human had time to analyze the playing style of the computer, and come up with a good counter-strategy. That is the main reason the computer lost.
Limiting the screen reduced the skill of the computer a few percentages, but it wasn't the only or (in my opinion) the main difference.
The difference between Jeopardy and the Starcraft AI is that the Starcraft AI wouldn't have come close to beating a human if it weren't for the inhuman precision.
The strategies the computer came up with were lousy (and very map-dependant), but it was able to compensate by having extremely precise movements.
We don't know exactly why Komincz won this game after losing the previous five
You could know if you'd watch the games. In the first set, DeepMind won with inhumanly superior micro. It was really cool, but computers have been better at micro for a long time. Speed and precision are things computers are good at, that's why we have aimbots.
In the second set, the human readjusted, and thought of strategies that would defend against the superior micro (by building more powerful units), while taking advantage of the computer's weaknesses (poor knowledge of army compositions, weak knowledge of positioning, and seemingly no object permanence: once enemy units are out of view, it has no idea where they are or if they exist).
Unless you believe there are only four continents "continent" being nonstandard. The Wikipedia page doesn't mention it, but some people consider Central America to be a separate continent, as well.
-- Insufferable Eurosnob
Shouldn't you be speaking French or something? True, in Spanish they call it futbol but they also say "pie" instead of foot, so I don't know what you're complaining about. Wait until you realize they call it "bano" instead of bathroom.
How did Siri respond? What was the joke?
It should be pointed out that by 1977, most of those had been invented.
Yeah, if you hire anyone (or look for a job in the Bay Area), you'll likely run into Hackerrank.
If Oracle can claim copyright on simply the function names of Java,
Because they've licensed it to the world, just like every other programming language worth using. Python APIs are just as copyright as Java's, but again, there is a license.
In the post-Oracle vs Google world, make sure you have a license for any APIs you are using.
Well well well. That's three deep subjects.
It's dead on the desktop
In fairness, almost everything is dead on the desktop at this point.
The thing I've recently heard about Java is that you are subject to Oracle's random whims. Right now, you can get and use the runtime environment and development environment for free, but you don't know if they will randomly decide to charge you a ton of money to use it and send an army of lawyers after you.
Use OpenJDK. Problem solved.
Between it's different versions, the security problems this brings, etc., it's dying fast in the professional environment
What security problems come with Java?
I don't see a lot of JSP servers, either
JSP is legacy now, but a lot of companies are using Java on the backend in a web services model. The frontend can be in Angular, or React, or whatever. It's a good choice when you want stable, cheap developers.
As my own professional opinion, I would say that Java is better for writing backend APIs than Node/Express.
Many things needed to go to the moon have not been used for a long time and hence are not available anymore.
Like what? Core memory? Guess we'll really regret not having that around anymore.
We do not have the capability to get people to the Moon
This is a ridiculous statement. We have the technology to go to the moon, technology has advanced since the '70s. It's just a matter of whether we want to do it or not.
That seems increasingly unlikely. The technology is there, it's just a matter of finding someone crazy enough to push it through and get it done.
That's a deep subject.
they've tricked a bunch of people into thinking that the flu shot is a vaccine, and that having everybody take it would create "herd immunity." Even though it clearly doesn't; the success rate is far too low to create that effect.
That's good to know, no more flu shot for me.
It's not a matter of smart or stupid here. Even a very smart person will come to the wrong conclusion if they have bad data.
When people are ignorant of statistics, they are unable to collect good information, and unable to figure out what is true among multiple interpretations. The correlation is real, what do they decide?
There is a very real correlation between venemous spiders and spelling bee champions. That is undeniable. Understanding the correlation takes a level of understanding of statistics. What kind of experiment would you develop to determine if spider bites are causing spelling bee champions?
For people who don't understand statistics, the world is a very confusing place, no matter how smart you are.
I have given up the fight against the autocomplete function on my smartphone already. There is no way to turn it off.
You can install an alternative keyboard that doesn't have auto-complete. If you want more specific advice as a shortcut, on Android, install the Korean keyboard (and of course only use the English half). It will give you suggestions in the top bar, but won't automatically complete anything for you unless you gesture for it.
If you are very picky about what you want the keyboard to do, then you can always build your own.
Have you tried this editor? What do you think of it? How is it?
The reason they used Python is because the most important use cases are already covered in Java or C++ with things like autocomplete. Adding AI doesn't give much improvement to what already exists. But since Python is dynamic, you can't know the type at compile time. That is why you need to use these extreme solutions for Python.
When was the last time you had an off-by-one error? It's not really very common.
What a brilliant counterargument. Intellectually stunning, really.