Artificial intelligence translates articles, the impact of artificial intelligence on translation

Arrente is flourished in the center of the business fulcrum in the Asahi red pepper and'Tricks'With three weights, we can give a score to each of the sticks of A, 1, 2, 3, 4, 4, 5, 6, 35, so that the first 4 questions are acres. I just didn't give a hint on how to get to that square root thingI was thinking that the possibility of requiring the fulcrum to be was was shifted from the center, so the weight of the sticks came into play Solution to (5):

I'll skip the explanation of how to mark the center point with respect to each stick and sharpen it 1,2 ((which is the solution to (1)--4)). To the front gives a standard weight to the 2nd point and along with the object with respect to the left does not know the weight x to balance it; The notation is balanced as yDotting this means xy=2

In relation to the left x art, point 1 expresses the weight that the object has and along with the positive side a standard weight balances it; Mark the balance points and of course, the points are xNow we have marked the dot y. on the left and x on the frontThe standard expression weight is characteristic y and a characteristic of the note is that x=y of x

conditions for those balanced. The rest is fairly easy: xy=2 and x=y=> x 2=2=>x= 2 ((1 2)).

(6) Solution: Balance the sticks in each aspect with a weight to give after ion 1There is a weight on the front of the stick about the left and balncing put three weights on the back ion 1 in relation to the front to give the back ion 3.

Place two plus the weight balance of Ate after ion 3 on the left and Ate after ion 1 to give after ion 7. I put a weight after the ion 1 on the left side and balance the sticks on the front of the object xtag as y

xy=1 balance position. Put object X on the back ion 7 and balance the weight that the stick has on the front. The equilibrium after the ion is related to the front is 7x

Put object x on the back ion 7x and balance the weight of the stick on the left side. The equilibrium posterior ions are about 7x weight after ion, 7x2 and one in yIn terms of conditions, plus the weight balance of the wood sticks, y=7x 2 places.

Combine it with xy=1 (1 3) to give x=(1 7).

At the fulcrum of the center, the juggling of the three weights, we can mark each side of the , paste5, m5, 45, 35 years old ......Hence the fourth question all.

I just don't know how to get to that square root thing. I think it may be necessary to shift the fulcrum, the approach of the center.

Resolution (5):

I'm going to skip explaining how to commemorate the center point, stickers on both sides of the point (this can be interpreted as solving (1)-(4).

Put a standard weight at 2 points on the right, balance it with the object and an unknown weight x on the left, marking the point for yThis means that the balance xy = 2.

Putting the object weight x point 1 on the left and balancing a standard weight on the right, Mark balances the point of view and, of course, the most critical is the x.

Now we have the X-rays on the left and right. Put a standard weight and a point y x point, if they are balanced, it means x = y.

The rest is very simple: xy = 2 and x = y = > x 2 = 2 = > x = 2 (1 2).

Solution: Balance insisted on one weight for one on each side. In placing three weight portions 1 on the left, in holding a weight balncing right position up 3 on the right. On the left side of the two parts 3 and weight, 1 part 7 pose.

A body weight in position 1 on the left and a balance sticking with object x on the right. The equilibrium position of Mark is y xy = 1. In the place of the object, X-ray is required to adhere to a weight balance, on the right.

The position of the balance is the group's current price expectation of 07 to the right. In the place of request object x and the balance group current price is expected to be 07 followed by a weight on the left. The position of the balance is to the left of the group's current price expectation of 07 2 Road.

One weight in one of the places, the current price of the group is expected to be 07 2 yIf the rod weight balance, y = group current price expected 07 2. Combine casually give x = = 1 10 (1) ().

The impact of AI on translation is both an opportunity and a challenge.

With AlphaGo's victory over world Go champion and professional Lee Sedol in a man-machine battle, people are starting to re-examine the potential of artificial intelligence. Before that, many people believed that Go represented the ultimate human intelligence, which could never be achieved by machines. But now, big data coupled with advanced algorithms has made mankind forever lose its pride in this "ultimate wisdom".

Although the AI represented by AlphaGo is not self-aware and does not "think like a human", it has the potential to do many jobs that currently seem to be only capable of the human brain. Is it possible for AI to replace humans in high-level translation? A professional translator gave the following words:

It's entirely possible. It's just not going to happen anytime soon.

Expand your knowledge:

At present, the strength of machine translation is still at the level of literal translation that can be understood, and in the eyes of professional translators, such a level is naturally unreachable, unproductive, and has little practical application value. Even when they see a poorly translated translation, professional translators will first wonder if it is "Google" translated. This has also reassured many translation practitioners that their work will not be replaced by machines one day.

However, if AI wants to reach or even surpass the level of professional translators, it can be achieved in other ways, such as big data analysis and high-level language interpretation algorithms, just as AlphaGo, which "can't think like a human", can beat humans in Go, which requires a lot of logical thinking. However, achieving high-quality translations in this way is still a huge challenge, which is why it is not possible in the short term.

Natural language processing.

What's involved: Natural language processing (NLP) is a field of computer science, artificial intelligence, linguistics that focuses on the interaction between computers and human (natural) language. Therefore, natural language processing is related to the field of human-computer interaction.

There are many challenges in natural language processing, including natural language understanding, and as a result, natural language processing involves an area of human-computer interaction.

Many of the challenges in NLP involve natural language understanding, where the computer originates from human or natural language input meaning, and others that involve natural language generation.

Modern NLP algorithms are based on machine learning, specifically statistical machine learning. The machine learning paradigm is different from the usual previous attempts at language processing. The implementation of a language processing task usually involves encoding a large set of rules directly by hand.

We have been talking a lot about artificial intelligence, and recently there is a research release that says that artificial intelligence is close to human translation in translation and other functions, and the following IT training will take a look at the specific situation.

The Microsoft team conducted multiple rounds of evaluation of the test set, each of which randomly selected hundreds of sentence translations. In order to verify that Microsoft's machine translation is as good as human translation, Microsoft did not stop at the requirements of the test set itself, but hired a group of bilingual language consultants from outside to compare Microsoft's translation results with human translation.

The complexity of the verification process is another reflection of the complexity of machine translation to be accurate. For other AI tasks, such as speech recognition, it's fairly simple to determine whether a system is performing like a human, because the ideal outcome is exactly the same for humans and machines, and researchers also refer to this task as a pattern recognition task.

Machine translation, however, is another type of AI task, where even two professional translators will have slightly different translations of the exact same sentence, and neither of them will be wrong. That's because there's more than one "correct" way to express the same sentence. Zhou Ming said:

That's why machine translation is much more complex than pure pattern recognition tasks, where people may use different words to mean exactly the same thing, but they may not be able to determine exactly which one is better. ”

Complexity makes machine translation a challenging problem, but it's also a very rewarding one. Liu Tieyan believes that we don't know when the machine translation system will be able to translate any language and any type of text, and can reach the level of professional translators in multiple dimensions such as "trust, reach, and elegance". However, he is optimistic about the progress of technology, because every year Microsoft's research team and the entire academic community invent a large number of new chain manuscript technologies, new models and new algorithms, "What we can do is that the application of new technologies will definitely improve the results of machine translation."

The research team also said that this technological breakthrough will be applied to Microsoft's commercial multilingual translation system products, which will help other languages or more complex and specialized texts to achieve more accurate and authentic translations. In addition, these new technologies can also be applied to other fields other than machine translation, leading to more breakthroughs in AI technology and applications.

Artificial intelligence is now getting more powerful and can even replace translation. But translation scholars also lose their jobs as a result. Where will they go from here?

In fact, I think that the work of translation scholars is more elaborate than that of artificial intelligence, and they can translate more professional and standard languages. Therefore, I think it is difficult to replace translation scholars in the short term. And in fact, with such a large amount of translation, artificial intelligence cannot be completed quickly, and it also needs the assistance of translation scholars.

In fact, translation scholars are professional translators, so they are usually very proficient in some professional terms, and even how to better translate another language into common usage. Since artificial intelligence is a computer, it is actually translated only according to the script, and cannot be translated into a really smooth or even very beautiful sentence, and there is no way for artificial intelligence to replace translation scholars in these aspects, just like foreign literary works generally require professional translation scholars to be able to translate the real connotation and beauty. Artificial intelligence paper can simply translate the meaning, but it will not pay attention to the beauty of words.

But if you do this, the translated literature is actually not good-looking, or even boring. Therefore, the translation of some literary works still needs professional translation scholars. <>

And in fact, in some professional terms, translation may be more powerful than artificial intelligence. For example, in some literature with a lot of professional terms, in fact, what artificial intelligence can do is only strong translation, but in fact, this requires a certain level of professionalism, such as some archaeological documents. These also need to be translated.

However, if it is translated by artificial intelligence, it may lose some of its meaning, and there may be no way to study archaeology. It has a propulsive effect, so it is still necessary for professional translation scholars to compare the dictionary with professional translation one by one. <>

And in fact, there are very, very many translation works in the world, and there are artificial intelligence that need a lot of translation every day, and there may be errors in their translations, after all, they are computers and machines, and they may pronounce them, and the sentences are not very smooth, or the words are not stacked beautifully. If it is not fluent, then it is also necessary for a translation scholar to proofread it. So I think the future work of translation scholars will also assist AI in translating and publishing.

These professional translation scholars can be used to proofread the errors or incoherence of the AI translation. So in the future, translation scholars will not be unemployed. They can form a new combination with artificial intelligence, which can speed up the efficiency of translation and make our translation work better and better.

I think that artificial intelligence cannot completely replace translation at present, after all, everyone speaks with different thinking logic, and some translation software cannot replace it, but human translation does have to consider the way back.

Artificial intelligence can only replace some simple translations, and it needs to be learned all the time, and its learning process is repeated training given by translation scholars. At the same time, in many important occasions, in fact, human translation is still required, emergency and fault tolerance.

Of course, you can continue to work, after all, machine translation still needs to be corrected manually, and you will not lose your job.

Translation scholars must also have a market demand, and they cannot be completely replaced, for example, in some high-end places, there is definitely a need for translation scholars to do translation work.