機器學習是否意味著理解科學的終結？
Could Machine Learning Mean the End of Understanding in Science?
Could Machine Learning Mean the End of Understanding in Science?
AUGUST 10, 2018
In January this year, scientists did just that. They used machine learning to accurately predict the outcome of a chaotic system over a much longer duration than had been thought possible. And the machine did that just by observing the system’s dynamics, without any knowledge of the underlying equations.
Awe, Fear, and Excitement
We’ve recently become accustomed to artificial intelligence’s (AI) dazzling displays of ability.
Last year, a program called AlphaZero taught itself the rules of chess from scratch in about a day, and then went on to beat the world’s best chess-playing programs. It also taught itself the game of Go from scratch and bettered the previous silicon champion, the algorithm AlphaGo Zero, which had itself mastered the game by trial and error after having been fed the rules.
Many of these algorithms begin with a blank slate of blissful ignorance, and rapidly build up their “knowledge” by observing a process or playing against themselves, improving at every step, thousands of steps each second. Their abilities have variously inspired feelings of awe, fear, and excitement, and we often hear these days about what havoc they may wreak upon humanity.
My concern here is simpler: I want to understand what AI means for the future of “understanding” in science.
If You Predict it Perfectly, Do You Understand It?
https://singularityhub.com/2018/08/10/could-machine-learning-mean-the-end-of-understanding-in-science/#sm.0001cvcvmh16oeeqpty4g2iim7gk4
一名學生痛擊拿下量子計算的其中一項頂級應用 - 現在怎辦？
A Student Took Down One of Quantum Computing’s Top Applications—Now What?
AUGUST 12, 2018
The possibility that quantum computing could turbocharge machine learning is one of the most tantalizing applications for the emerging technology. But an 18-year-old student just put that vision in doubt after finding a classical solution to one of its most promising real-world applications.
One of the poster boys for the incipient field of quantum machine learning (QML) was a solution to the “recommendation problem”—essentially, how Netflix determines what movie you might like—published in 2016 that was exponentially faster than any classical algorithm. But as reported in Quanta, in the process of trying to verify its unassailability, Ewin Tang came up with a classical version just as fast.
Tang, a prodigy who enrolled at the University of Texas at age 14, was set the task of proving there was no fast classical alternative to the quantum solution by quantum computing expert Scott Aaronson as an independent research project. Ultimately, he ended up taking inspiration from the quantum algorithm to design a classical one exponentially faster than any of its predecessors.
His research is now undergoing peer review before publication, but has already been presented informally at a meeting of quantum computing experts who broadly agreed that the algorithm was correct, according to Quanta.
In some senses it’s good news, because Tang has found a way to translate speedups thought to only be possible with an (as yet unbuilt) quantum computer to more conventional machines. But that’s probably little comfort if you’re one of the companies investing millions trying to build quantum hardware.
Machine learning has long been considered one of the potential early applications for quantum computers, because there are fundamental synergies between the two fields. They’re both best when dealing with huge amounts of data; they’re resistant to uncertainty; and they can tease out subtle patterns traditional computing approaches may miss. The hope is that large-scale quantum machines will ultimately be able to perform these calculations exponentially faster than machine learning running on conventional hardware.
https://singularityhub.com/2018/08/12/a-student-took-down-one-of-quantum-computings-top-applications-now-what/#sm.0001cvcvmh16oeeqpty4g2iim7gk4
One of the poster boys for the incipient field of quantum machine learning (QML) was a solution to the “recommendation problem”—essentially, how Netflix determines what movie you might like—published in 2016 that was exponentially faster than any classical algorithm. But as reported in Quanta, in the process of trying to verify its unassailability, Ewin Tang came up with a classical version just as fast.
Tang, a prodigy who enrolled at the University of Texas at age 14, was set the task of proving there was no fast classical alternative to the quantum solution by quantum computing expert Scott Aaronson as an independent research project. Ultimately, he ended up taking inspiration from the quantum algorithm to design a classical one exponentially faster than any of its predecessors.
His research is now undergoing peer review before publication, but has already been presented informally at a meeting of quantum computing experts who broadly agreed that the algorithm was correct, according to Quanta.
In some senses it’s good news, because Tang has found a way to translate speedups thought to only be possible with an (as yet unbuilt) quantum computer to more conventional machines. But that’s probably little comfort if you’re one of the companies investing millions trying to build quantum hardware.
Machine learning has long been considered one of the potential early applications for quantum computers, because there are fundamental synergies between the two fields. They’re both best when dealing with huge amounts of data; they’re resistant to uncertainty; and they can tease out subtle patterns traditional computing approaches may miss. The hope is that large-scale quantum machines will ultimately be able to perform these calculations exponentially faster than machine learning running on conventional hardware.
https://singularityhub.com/2018/08/12/a-student-took-down-one-of-quantum-computings-top-applications-now-what/#sm.0001cvcvmh16oeeqpty4g2iim7gk4
沒有留言:
發佈留言