Defu human-machine war ended, human defeat, artificial intelligence capital drama continues

After five days of fighting, Libratus, the Texas Hold'em AI system, won the final victory without any suspense. In this Texas Hold'em man-machine battle, the six players of the Chinese Dragon Team played a total of 36,000 hands with the Cold Fight Master, losing a total of 792,327 points, and AI beat humans. The outcome of the game was not unexpected. Before the start of the competition, Li Kaifu, founder and chairman of the Innovation Workshop, said: "I feel pessimistic about human beings, and the probability of winning is less than 10%."

In fact, in January of this year, at the Rivers Casino in Pittsburgh, Pennsylvania, the Libratus artificial intelligence system developed by CMU defeated the top human professional poker players. The game lasted for 20 days, playing 120,000 hands, and finally AI. Won $1.76 million.

Defu human-machine war ended in human defeat, artificial intelligence capital drama continues

Bluffing AI

Libratus is an artificial intelligence program for playing unlimited Texas Hold'em. Libratus' strategy is not based on the experience of professional players, so its playing style may be significantly different. Based on approximately 15 million core hours of calculations at the Pittsburgh Supercomputer Center, it uses algorithms to analyze Texas Hold'em rules to build its own strategy, and it can think about its next step in the game by predicting the winning percentage of all future steps.

Unlike Go, which emphasizes computational and formal judgment, Texas Hold'em is more about the multiplayer process, how to avoid weaknesses such as human greed, and a good match between scientific probability statistics and flexible combat strategies. In games such as Go and Chess, machines and humans can get all the information before making a decision. In Texas Hold'em, they can't know each other's cards and don't know what the next card is issued by the dealer. Under "incomplete information," artificial intelligence needs to guess knowledge about the opponent's card and the next card based on empirical or probabilistic knowledge, and then develop its own coping strategies.

"If AlphaGo is a super genius, the CMO system is actually an EQ expert to some extent, and it is EQ to defeat you." Kai-fu Lee said. In order to reduce the luck factor in the licensing game, the robot adopts the compound symmetric licensing, and one of the two pairs of players will get the robot card with the paired players, so the six players will be split into two. The room is played against the master of the cold, and the players must be sure that they can't meet each other during the game. Throughout the game, the masters of the cold puff also often have some "singular" styles, such as frequent betting on the pool, causing great pressure on the opponent, and making bluffs that humans can't do for psychological reasons but correct. fool).

"AI uses enhanced learning technology to learn the best poker gameplay from the self-game, and avoid learning from the established model of human beings. This is very important." Kai-fu Lee said. However, it is understood that Libratus's algorithm is currently only applicable to one-on-one competitions with unlimited bets. If the game is extended to the more common multiplayer competition, Libratus will face more challenges and need to be strategically Upgrade and adjustment.

Artificial intelligence application challenge

Like Ma Yun’s question, artificial intelligence defeats the master of human Go, so what is the value of artificial intelligence that plays poker in solving a wider range of real problems? In Kai-Fu Lee's view, most of the information in the world is not yet public. The cold-blooding master's ability to reason in the face of incomplete or misleading information can solve uncertainties in decision-making, diplomacy, business cooperation, and negotiation in the future. Become a "personnel" of humanity.

But as Google's artificial intelligence beats the master of human Go, AlphaGo's ability is fully recognized, but it is not obvious in terms of commercialization methods and capabilities. Taking AlphaGo as an example, you need to learn a large number of chess games to master effective chess skills. The current form of the cold-fighting masters is one-on-one. In real life, the German game is a multiplayer game, and the multiplayer game is calculated. The complexity is not currently available to the masters of the cold.

How artificial intelligence technology extracted by artificial intelligence and cold-float masters can exert its commercial value. Li Kaifu believes that three conditions still need to be met: massive data, data labeling, and single field. For example, in the financial sector, finance is virtual and created by people. The amount of data is huge and it is inherently labeled. For example, whether the stock’s daily limit, whether the microloan is repaid, or whether it is accidental after buying insurance is an indication. AI has potential in lending, banking, investment, and insurance.

"Compared to telling people what artificial intelligence can do, it is more important now to tell people that artificial intelligence can't do anything." Yu Kai, founder and CEO of Horizon Robotics, told the First Financial Reporter that he was watching Come, the primary challenge of further expansion of artificial intelligence is the problem of insufficient data. As we all know, artificial intelligence is based on massive data, and the algorithm model is optimized through big data training. Taking face recognition technology as an example, training this algorithm model requires at least one million levels of image data.

At present, artificial intelligence is mainly supervised learning, and supervised training requires tagged data, so the quality and accuracy of the data are closely related to the output. “How to eliminate noise and spam in the data and get high quality and tagged data becomes a new challenge, which involves unsupervised learning or semi-supervised learning.” Horizon Robotics co-founder and vice president of algorithms Huang Say it.

Another major challenge is the lack of deep learning and the ability to migrate scenarios. Data in each area needs to be re-collected, standardized, and retrained, making it difficult to promote across domains. These challenges are also an urgent need for breakthroughs in the artificial intelligence industry and academia. “In the process of hiring, there are many people who know how to study deeply, and there are very few people who know how to migrate and learn, and have the ability to think.” The fourth paradigm founder and CEO Dai Wenyuan told reporters.

At the practical level, artificial intelligence still has a long way to go. Just as Li Kaifu responded after the game: "Artificial intelligence has extended from the perfect information of AlphaGo to the cold-fighting master of imperfect information. There is basically no suspense in man-machine battle. It is reported that AlphaGo is coming to China and Kejie in the near future. It is no longer of scientific significance. In the future, it should pay more attention to the artificial intelligence in the commercial field and the commercial value generated in the fields of finance, medical care and education.

The artificial intelligence capital drama continues to be staged: investment institutions are crazy

Artificial intelligence has become the main direction after the Internet enters the "second half". The giants are willing to pay great attention to the gold, and the capital market is even more reluctant to miss the enthusiasm. Li Kaifu, founder and chairman of the Innovation Workshop, recently spoke for the AI. "I am not Kai-Fu Lee, I am artificial intelligence." Zhenge Fund, IDG and other frequent shots, Vientiane Artificial Intelligence Research Institute, Lenovo Star's global artificial intelligence incubator Comet Labs, Junyi incubator artificial intelligence industrial park and other windward follow-up. Capital is performing an artificial intelligence drama.

Newborn darling

In the future, the innovation workshop will be turned into an 'AI factory', and 50% of the investment has been put in it. "Li Kaifu has become an evangelist of artificial intelligence, and innovative factories have indeed worked hard in the field of artificial intelligence. Recently, the Innovation Workshop and the Hainan Eco-Software Park jointly launched the first German-style human-machine war.

According to statistics, the Innovation Workshop has invested in nearly 30 startup companies that directly use artificial intelligence technology to create commercial value, including the contempt technology with face recognition technology face++, the unmanned technology company, the technology, and the artificial intelligence technology service provider. Four paradigm, chip technology company Horizon robot, family companion intelligent hardware equipment company Xiaoyuer Technology, American programming education robot company fantasy workshop.

These companies have taken the lead in commercial products. In addition to these companies that have become climatic in the commercial sector, in 2016, around the development strategy of “Deep Incubation + Regional Incubation + Business School”, the Innovation Factory also established the Shenzhen Incubation Platform, and established the Artificial Intelligence Engineering Institute, which also established business school operations. The model, the way of running through the traditional business school, has reached the cooperation with two leading universities.

More than just innovative workshops in the wind, Zhenge Fund, IDG, etc. are also frequently shot, and the areas involved in investment include all aspects of artificial intelligence such as driverlessness, intelligent hardware, and voice interaction.

Enthusiastic is also listed companies. Recently, the Vientiane Artificial Intelligence Research Institute jointly launched by the three listed companies, namely Dongfang Netpower, Jingshan Light Machine and Tomson Bianjian, was formally established. It plans to invest no more than 500 million yuan to set up research and research in Beijing, Wuhan, Shenzhen and Silicon Valley. The incubator, with reference to the innovative nature of the fund model operation, helps the research results to bear fruit in multiple industries and across industries with a diversified and open structure.

Taste the sweetness

According to the "2017-2022 China Artificial Intelligence Industry Market Prospect and Investment Strategic Planning Analysis Report" published by Prospective Industry Research Institute, in 2014, China's artificial intelligence industry market size was 4.86 billion yuan; as of the end of 2016, the artificial intelligence industry market The scale has grown to 9.56 billion yuan, with an average compound annual growth rate of 40.25%. With the expansion of the scope of artificial intelligence applications, it will drive the rapid growth of the industry scale. It is estimated that by 2018, the market size of China's artificial intelligence industry will reach 20.33 billion yuan.

Behind the surge in scale is the help of the giants, the cheers of entrepreneurs, and the joy of capital. In fact, entrepreneurs often have the technical research strength, lack of attention to the commercial level, and will not explore application scenarios. Investors, though not necessarily well-informed, are hungry for finding promising entrepreneurial projects and assisting in commercial development.

According to the 2016 annual report released by the Innovation Workshop, the innovation workshop achieved operating income of 85.42 million yuan in 2016, a year-on-year increase of 69.28%. Among them, the rapid growth of entrepreneurial service income and fund management outsourcing service income is the main reason for the increase in the company's operating income.

Senior investors in the investment industry said that some practices in the domestic market have always referred to the United States, and the giants of the United States have long invested heavily in artificial intelligence to make this market very hot. "VC invests in the big market, the Internet, and the mobile Internet. After the window of demographic dividend is closed, at present, artificial intelligence is the most likely big opportunity to improve the efficiency of the whole industry through artificial intelligence, even more than humans do something. Have the opportunity to ".

Long distance race

"Innovative workshops have a group of people who are most familiar with AI, and there are many AI projects that are being hatched. If anyone wants to invest in artificial intelligence, it is enough to innovate the workshop." Promote the advantages of artificial intelligence and innovative workshops in this area. It became a topic that Li Kaifu must talk about every time he attended various events.

In fact, artificial intelligence mainly depends on the team, and experts in the vertical field are the most sought after. The capital market does not want to pursue the fake wind, but also requires real technology, and hopes to grab the top talents. This has made the corners of the capital more and more fierce.

Liu Guang, the founder of the Vientiane Institute and the chairman of Dongfang Netpower, believes that first-class talents will often bring first-class projects. Vientiane's core strengths are openness, willingness to pay, data, manpower and time to serve talents. The judging criteria for technical talents will be a long-term achievement, including the value of the product, the company, and the impact on the entire industry.

Li Jian, chairman of Jingshan Light Machine, said that ordinary investment institutions cannot provide technical personnel with the possibility of diversifying with the industry. Vientiane Research Institute will truly operate enterprises, provide products, and serve the market, not just investing. There is an essential difference between the two.

The above-mentioned investment industry veterans believe that the current situation of AI is unclear, the capital is not in a hurry, the layout may be backward, and the layout is afraid of making mistakes. As a result, more and more discussions have been generated, and even long-term competition has arisen. "After all, the opportunity for entrepreneurs to seize a good field is not bad. This market is big enough and the barriers are strong, so the startup company will not do general artificial intelligence. The advantage of the giant is that it can still be invested even if it does not wading itself. Occupy technology, capital is as much as possible to seize the dominant resources," he analyzed.

Pooling Pumps

Pooling Pumps,Dc Water Pump,High Pressure Water Pump,Self-Circulation Water Pump

Sensen Group Co., Ltd.   , https://www.sunsunaquariums.com

This entry was posted in on