AlphaGo announced that he will fight again for the human honour of Go.
In March of this year, AlphaGo and Li Shishi's Go matchup, like the human fleet in the science fiction "Three-body", faced the first drop of water. This man from the letter is full of unpredictable and powerful combat power and has been caught off guard. In the "Three-body", the human team survived four warships, and Li Shishi, who was mad at AlphaGo for four innings, also saved a little face for mankind in the fourth inning of the five-game battle. After the Deep Mind team appeared on the Blizzard Carnival on the weekend, and announced that AlphaGo will challenge StarCraft II, this morning, AlphaGo team member Fan Wei announced on Weibo: AlphaGo's chess has made tremendous progress in the past six months and will return to play in early 2017. The AlphaGo team will announce more in the near future. AlphaGo shot again, the goal may be pointing to the world number one Currently in the player's points ranking, in front of AlphaGo, only Chinese chess player Ke Jie. In fact, during AlphaGo's Li Shishi battle, Ke Jie's confident remarks on Weibo have already made him a fire. "Even if the Alpha dog defeated Li Shishi, but it won't win me," "Do you know whether you are an Alpha dog or an Alpha cat! I haven't seen a big wind on the board of Ke Jie? Let the storm come back violently!" The "extraordinary" remarks during the "human-machine war" directly led Ke Jie to the cusp of public opinion. Although in July this year, due to the continuous loss to two lower ranked players, Ke Jie temporarily gave the world's first throne for two years to AlphaGo, but in the current rankings, since AlphaGo is no longer In the open battle, Ke Jie's points returned to the first position. The match at the beginning of next year is likely to be the first real matchup in the world: Ke Jie vs. AlphaGo. In fact, probably in June of this year, Yang Junan, secretary of the Party Committee of the National Sports General Administration's chess and card sports management center, revealed at the 37th Wuxi World Go Amateur Championship game briefing and draw ceremony: In the previous paragraph, the Chinese Go Association and Google Company had several contacts. Both parties agreed to arrange one of the top-ranked Go players Ke Jie and AlphaGo to perform the "human-machine final match" in the near future. Of course, whether it will be against Kee in AlphaGo in 2017 will be confirmed by the official announcement of the AlphaGo team. Looking at the future, what role might artificial intelligence like AlphaGo play in the future? Ultimately, it is the deep neural network that supports AlphaGo to win. In addition to Go and StarCraft II, the application of competitive games, when Google uses deep neural networks in translation, it can make translations smoother and clearer. According to Google, the new translation system is referred to as GoogleNeuralMachine Translation (GNMT), and in some languages, the GNMT error rate can be reduced by 60%. (Comprehensiveness of human translation, neural machine translation systems, and phrase word translation systems (to a degree of 0 to 6) in different language applications) Mike Schuster, an engineer at the project and a member of the Google think tank, said: We will train the entire system in an "end-to-end" model, which will allow the company to focus more on reducing the error rate of translation. Although it is not perfect now, it is indeed a lot better. (How does the new neural machine translation system detect a sentence and consider the relevance and importance between it) In machine translation, the deep neural network used by Google is called LSTM (longshort-termmemory – long-term short-term memory). LSTM retains long-term and short-term memory, a bit like the human brain, which allows it to learn in more complex ways. Unlike Google's previous phrase translation system (Phrase-BasedMachineTranslation-PBMT), LSTM can remember a sentence from beginning to end and then translate it, so that long sentence translation is no longer a patchwork of words, but a smooth and fluent language. In fact, for many years, Google has always wanted LSTM to serve translation, but it has not been able to overcome the problem of translation speed. It was not until a lot of engineering and algorithmic work was done that the model was perfected to allow the new neural machine translation system to keep working while spanning large Internet services. In addition to translation, as artificial intelligence and machine learning similar to AlphaGo become more intelligent, we will face the reality that machines will gradually replace human work. As Tesla's ElonMusk said in an interview with CNBC: Machines equipped with artificial intelligence are entering the labor market. For humans, this is likely to mean job loss and "generalized income." Because automation is the norm, our employment options may be limited in the future, but this will give people more time to enjoy their lives. In addition to the human translation mentioned above, the fastest candidate to be replaced is the car driver. Although most manufacturers now only provide driving assistance functions that meet SAE and NHTSALevel2. However, from the current trend of autonomous driving, the official civil use of fully automated vehicles may be only five or six years away from us, and the key to this is the driverless algorithm. In general, the current future of machine learning and artificial intelligence represented by AlphaGo is immeasurable. Of course, as the author of the love fan, Zeng Qingwei said: The best times are the worst times. With the help of machine learning, humans have the ability to study complex problems that traditional sciences cannot study; people can even deploy models with unclear mechanisms directly into products to help humans solve problems. However, with the rapid rise of artificial intelligence, the employment structure of the society has also encountered a huge test. The market for simple human work is bound to be shrunk or transferred by the squeeze of robots and artificial intelligence. For researchers and companies in the field of machine learning, this is the best time; for others, perhaps a job crisis is ambushing in the near future. Sun Kee Envelopes Limited , https://www.sunkeeenvelope.com
