高考英语外刊 人工智能将思想转化文字.docx

1、高考英语外刊 人工智能将思想转化文字Mind-reading AI turns thoughts into words using a brain implant30 March 2020ByJason Arunn MurugesuMind-reading AI turns thoughts into words using a brain implantAn artificial intelligence canaccurately translate thoughts into sentences, at least for a limited vocabulary of 250 word

2、s. The system may bring us a step closer to restoring speech to people who have lost the ability because of paralysis.Joseph Makin at the University of California, San Francisco, and his colleagues used deep learning algorithms to study the brain signals of four women as they spoke. The women, who a

3、ll haveepilepsy, already had electrodes attached to their brainsto monitor seizures.Each woman was asked to read aloud from a set of sentences as the team measured brain activity. The largest group of sentences contained 250 unique words.The team fed this brain activity to a neural network algorithm

4、, training it to identify regularly occurring patterns that could belinked to repeated aspects ofspeech, such as vowels or consonants. These patterns werethen fed to a second neural network, which tried to turnthem into words to form a sentence.Each woman repeated the sentences at least twice, and t

5、he final repetition didnt formpart ofthe training data, allowingthe researchers to test the system.Each time a person speaks the same sentence, the brain activity associated will be similar but not identical. “Memorising the brain activity of these sentences wouldnt help, so the network instead has

6、to learn whats similar about them so that it can generalise to this final example,” says Makin. Across the four women, the AIs best performance was an average translation error rate of 3 per cent.Makin says that using a small number of sentences made it easier for the AI to learn which words tend to

7、 follow others.Forexample, the AI was able to decode that the word “Turner” wasalways likely to follow the word “Tina” in this set of sentences, from brain activity alone.The team tried decoding the brain signal data into individual words at a time, rather than whole sentences, but this increased th

8、e error rate to 38 per cent even for the best performance. “So the network clearly is learning facts about which words go together, and not just which neural activity maps to which words,” says Makin.This will make it hard toscale up the system to a larger vocabulary because each new word increases

9、the number of possible sentences, reducing accuracy.Makin says 250 words could still be useful for people who cant talk. “We want to deploy this in a patient with an actual speech disability,” he says, although it is possible their brain activity may be different from that of the women in this study

10、, making this more difficult.Sophie Scott at University College London says we are a long way from being able to translate brain signal data comprehensively. “You probably know around 350,000 words, so its still an incredibly restricted set of speech that theyre using,” she says.（481）（一、外刊阅读：完型填空二、M

11、ind-reading AI turns thoughts into words using a brain implantAn artificial intelligence canaccurately translate thoughts into sentences, at least for a limited vocabulary of 250 words. The system may bring us a step closer to （41）_ speech to people who have lost the ability because of paralysis.Jos

12、eph Makin at the University of California, San Francisco, and his colleagues used deep learning algorithms to study the brain (42) _ of four women as they spoke. The women, who all haveepilepsy, already had electrodes attached to their brainsto (43) _ seizures.Each woman was asked to read aloud from

13、 a set of sentences as the team measured brain activity. The largest group of sentences (44) _ 250 unique words.The team fed this brain activity to a neural network algorithm, training it to identify regularly (45) _ patterns that could belinked to repeated aspects ofspeech, such as vowels or conson

14、ants. These patterns werethen fed to a second neural network, which tried to turnthem into words to (46) _ a sentence.Each woman repeated the sentences at least twice, and the final repetition didnt formpart ofthe training data, (47) _the researchers to test the system.Each time a person speaks the

15、same sentence, the brain activity associated will be similar but not identical. “Memorising the brain activity of these sentences wouldnt help, so the network instead has to learn whats similar about them so that it can generalise to this final example,” says Makin. Across the four women, the AIs be

16、st (48) _ was an average translation error rate of 3 per cent.Makin says that using a small number of sentences made it easier for the AI to learn which words tend to follow others.Forexample, the AI was able to decode that the word “Turner” wasalways likely to follow the word “Tina” in this set of

17、sentences, from brain (49) _ alone.The team tried decoding the brain signal data into (50) _ words at a time, rather than whole sentences, but this increased the error rate to 38 per cent even for the best performance. “So the network clearly is learning facts about which words go together, and not

18、just which neural activity (51) _to which words,” says Makin.This will make it hard to (52) _ the system to a larger vocabulary because each new word increases the number of possible sentences, reducing (53) _.Makin says 250 words could still be useful for people who cant talk. “We want to deploy th

19、is in a patient with an actual speech disability,” he says, although it is possible their brain activity may be different from that of the women in this study, making this more (54) _.Sophie Scott at University College London says we are a long way from being able to translate brain signal data comp

20、rehensively. “You probably know around 350,000 words, so its still an incredibly (55) _ set of speech that theyre using,” she says.（481）41.A. inspectingB. restoringC. admiring D. inspiring42.A. emotion B. attractivenessC. awarenessD. signals43.A. monitorB. master C. controlD. expect44.A. concludedB.

21、 excludedC. containedD. increased45.A. extended B. occurringC. ignoredD. concerned46.A. formB. handle C. handD. force47.A. issuing B. producingC. allowingD. acquiring48.A. behaviorB. commentC. preparationD. performance49.A. possibility B. activityC. capacityD. responsibility50.A. individualB. financ

22、ial C. social D. technical51.A. servesB. finishes C. mapsD. competes52.A. switch upB. put upC. rise upD. scale up53.A. privacyB. accuracyC. currencyD. fluency54.A. critical B. specific C. properD. difficult55.A. committedB. oppressedC. restrictedD. dominated 二、参考答案BDACB ACDBA CDBDC三、原文链接四、核心词汇accura

23、te accuracies accuracy accurately inaccuracies inaccuracy inaccurate inaccurately acquire acquired acquirer acquirers acquires acquiring unacquired attract attracted attracting attraction attractions attractive attractively attractiveness attractor attractors attracts unattractive unattractively awa

24、re awareness unaware unawares commit commitment commitments commits committal committals committed committing uncommitted critic critical critically critics uncritical uncritically currency currencies dominate dominated dominates dominating domination extend extendable extended extender extenders ex

25、tending extends unextended general generalisabilty generalisable generalisation generalisations generalise generalised generalises generalising generalist generalists generalities generality generalizabilty generalizable generalization generalizations generalize generalized generalizes generalizing

26、generally ignore ignored ignores ignoring inspect insp inspected inspecting inspection inspections inspector inspectors inspects inspire inspiration inspirational inspirations inspired inspires inspiring uninspired uninspiring memory memorial memorials memories memorisation memorisations memorise me

27、morised memorises memorising memorization memorizations memorize memorized memorizes memorizing occur occurred occurrence occurrences occurring occurs reoccur reoccurred reoccurring reoccurs oppress oppressed oppresses oppressing oppression oppressions oppressive oppressively oppressiveness oppresso

28、r oppressors restore restoration restorations restorative restored restorer restorers restores restoring unrestored restrict restricted restricting restriction restrictions restrictive restrictively restricts unrestricted unrestrictive scale scalable scaled scales scaling scalings signal signaling s

29、ignalled signalling signally signals五、原文翻译Mind-reading AI turns thoughts into words using a brain implant读心术人工智能通过植入大脑将思想转化为语言An artificial intelligence canaccurately translate thoughts into sentences, at least for a limited vocabulary of 250 words. The system may bring us a step closer to restoring

30、 speech to people who have lost the ability because of paralysis.人工智能能够准确地将思想翻译成句子，至少在250个单词的有限词汇范围内。这个系统可能会使我们更接近于恢复那些因瘫痪而丧失语言能力的人的语言能力。Joseph Makin at the University of California, San Francisco, and his colleagues used deep learning algorithms to study the brain signals of four women as they spoke. The women, who all haveepilepsy, already