The laws of thermodynamics: the pedant's best friend.
For actual engineering and public policy decisions, the renewable/non-renewable categorization, as it is conventionally understood, is both meaningful and useful.
It's called "renewable" here. Excluding hydro, it generated 2% of the national total in 2015 and 7% in 2016, about the same as coal. I couldn't find more recent numbers unfortunately. In 2016 there was one province that was 98% wind.
Nah, they'll block the obvious sites. They're responding to a bad press attack by a special interest group. They need to do something so they can say, see, we did something.
Sites that serve ads are held responsible for damages if visitors get hijacked by those ads. In turn, those sites can hold ad providers liable. The online advertisers would tighten up their security in a hurry when the lawsuits started rolling in. We might even get to go back to plain image ads.
No, it can't. The human neural network processes a large amount of data, not just vision but lots of other sensory input as well, and it does it a lot faster than any existing computer doing anything of even vaguely comparable sophistication. We don't know exactly how 30 years of a human observing and learning about the world translates into analogously training an artificial neural network, but it is definitely more than a weekend's worth. Reasonable estimates put it at more than 30 years worth, not to mention using a truly enormous amount of training data. One of the big problems with reinforcement learning is that nobody wants to wait long enough for it to work in the real world so systems are trained in vastly simplified simulations (like 80s video games). When you vastly simplify, you necessarily lose things.
I have some experience with this. I very strongly suspect you do not. But you don't even have to believe me. Geoff Hinton has made the same argument, and he has some pretty reasonable qualifications.
It develops when false positives didn't hurt as much as false negatives, over an evolutionarily relevant timescale. Cognitive biases can screw you in new situations, and don't operate well at at the group level.
It would be interesting to compare the ethnic imbalance in prison populations between China, about midlevel in incarceration rates in the world, and the USA, the unchallenged world leader in the field. The issue is well studied in the US, where there is very significant ethnic imbalance. I couldn't find data for China.
US male life expectancy at 65 has been holding fairly steady at 17.9 years. So if that 81 figure is correct, the average male takes out more than he put in. Women get almost an extra three years.
Mean (not median) household income in the US in 2014 was $72,641. It depends what you mean by "current standard of living" of course, but 72k sounds okay. Mean income was above median income, and median would be a reasonable way of defining "current standard of living."
The targets make some allowances for the fact that that US emits more CO2 per capita than anybody except some middle eastern oil nations, a few Caribbean islands and Luxembourg (for some reason), and many times as much as many.
But the cost of shipping by train does not always reflect the actual cost. Apparently the rail industry is the subject of some rather unfair tax policies due to it being an absolute cash cow in the past.
You're correct. China's homicide rate is MUCH lower than in the US. There's a much lower imprisonment rate in China too. China has a healthy lead on the US in the capital punishment rate, but in either country it's just a rounding error against imprisonment or homicide.
Reinforcement learning basically exists to solve problems that have the properties you describe. Researchers in the field have been aware of them for a long time. Many modern reinforcement learning algorithms basically use artificial neural networks to estimate the trickier bits in a Q-learning framework. Q-learning was introduced in 1989 and the basic theory developed in the early nineties.
Humans are not slow. Hinton has computed the amount of sensory information that is processed by the human brain, using reasonable approximations for things like the effective sampling rate of the eyes and ears. It's enormous.
The *consciousness* that we subjectively experience is slow. We're also pretty horrible at tasks we have to consciously think about as we're doing them too. Both of which suggest that "consciousness" might be considerably less important than many give it credit for.
That is an experiment that's being done. The idea is to learn how much behaviour can arise spontaneously with a reinforcement learning algorithm, some basic motivations, and real-world sensory input.
That IS how we do a great deal of our learning, almost exclusively when we're young. When we advance a little further we get some direct supervised learning mixed in.
Slashdot Mirror
The laws of thermodynamics: the pedant's best friend.
For actual engineering and public policy decisions, the renewable/non-renewable categorization, as it is conventionally understood, is both meaningful and useful.
"By the time that runs out"
That phrase is the key to why fission isn't considered a renewable source.
I'm curious whether that's actually true.
It's certainly false if you compute the rate per kWh for all power use, including that used to produce food and for life support.
But in absolute terms? How much solar-related skin cancer is there, versus cancer that could be attributed to air and water pollution?
It's called "renewable" here. Excluding hydro, it generated 2% of the national total in 2015 and 7% in 2016, about the same as coal. I couldn't find more recent numbers unfortunately. In 2016 there was one province that was 98% wind.
Decent chunk, and growing fast.
Well, Starbucks is blocking the porn industry and YouPorn apparently blocked Starbucks. Does that count?
Nah, they'll block the obvious sites. They're responding to a bad press attack by a special interest group. They need to do something so they can say, see, we did something.
Sites that serve ads are held responsible for damages if visitors get hijacked by those ads. In turn, those sites can hold ad providers liable. The online advertisers would tighten up their security in a hurry when the lawsuits started rolling in. We might even get to go back to plain image ads.
Hornwumpus: "there are no self-driving cars"
Evidence to support: CEO says a self-driving car that can handle "all conceivable scenarios" is impossible. At least, any time soon.
The FUD is strong in this one.
Google (or Waymo) might beg to differ.
Perhaps you meant "there are no self driving cars THAT YOU CAN BUY today."
In 2050 Putin will be 98 years old, so I'm not so sure how mysterious it would be if he died before then.
It's even funnier that the summary's grammatical error accidentally describes reality.
Meta-irony, if you will.
Trees are made of carbon sucked out of the atmosphere. Burying trees is an effective carbon sequestration strategy, if you bury them deep enough.
No, it can't. The human neural network processes a large amount of data, not just vision but lots of other sensory input as well, and it does it a lot faster than any existing computer doing anything of even vaguely comparable sophistication. We don't know exactly how 30 years of a human observing and learning about the world translates into analogously training an artificial neural network, but it is definitely more than a weekend's worth. Reasonable estimates put it at more than 30 years worth, not to mention using a truly enormous amount of training data. One of the big problems with reinforcement learning is that nobody wants to wait long enough for it to work in the real world so systems are trained in vastly simplified simulations (like 80s video games). When you vastly simplify, you necessarily lose things.
I have some experience with this. I very strongly suspect you do not. But you don't even have to believe me. Geoff Hinton has made the same argument, and he has some pretty reasonable qualifications.
It develops when false positives didn't hurt as much as false negatives, over an evolutionarily relevant timescale. Cognitive biases can screw you in new situations, and don't operate well at at the group level.
https://arxiv.org/abs/1802.102...
It would be interesting to compare the ethnic imbalance in prison populations between China, about midlevel in incarceration rates in the world, and the USA, the unchallenged world leader in the field. The issue is well studied in the US, where there is very significant ethnic imbalance. I couldn't find data for China.
US male life expectancy at 65 has been holding fairly steady at 17.9 years. So if that 81 figure is correct, the average male takes out more than he put in. Women get almost an extra three years.
Mean (not median) household income in the US in 2014 was $72,641. It depends what you mean by "current standard of living" of course, but 72k sounds okay. Mean income was above median income, and median would be a reasonable way of defining "current standard of living."
The targets make some allowances for the fact that that US emits more CO2 per capita than anybody except some middle eastern oil nations, a few Caribbean islands and Luxembourg (for some reason), and many times as much as many.
But the cost of shipping by train does not always reflect the actual cost. Apparently the rail industry is the subject of some rather unfair tax policies due to it being an absolute cash cow in the past.
Ah, the cause of so much strife in the world:
1. Describe some perfectly reasonable/believable/true observation.
2. Come up with some crazy and unsupported statement that (1) supports some claim.
3. Profit?
You're correct. China's homicide rate is MUCH lower than in the US. There's a much lower imprisonment rate in China too. China has a healthy lead on the US in the capital punishment rate, but in either country it's just a rounding error against imprisonment or homicide.
Reinforcement learning basically exists to solve problems that have the properties you describe. Researchers in the field have been aware of them for a long time. Many modern reinforcement learning algorithms basically use artificial neural networks to estimate the trickier bits in a Q-learning framework. Q-learning was introduced in 1989 and the basic theory developed in the early nineties.
Humans are not slow. Hinton has computed the amount of sensory information that is processed by the human brain, using reasonable approximations for things like the effective sampling rate of the eyes and ears. It's enormous.
The *consciousness* that we subjectively experience is slow. We're also pretty horrible at tasks we have to consciously think about as we're doing them too. Both of which suggest that "consciousness" might be considerably less important than many give it credit for.
That is an experiment that's being done. The idea is to learn how much behaviour can arise spontaneously with a reinforcement learning algorithm, some basic motivations, and real-world sensory input.
That IS how we do a great deal of our learning, almost exclusively when we're young. When we advance a little further we get some direct supervised learning mixed in.