Netease Technology News July 15 news, organized by the Netease media theme for the "new" 2017 Netease future technology summit today in Beijing. Stanford University tenured professor and academician of the American Academy of Sciences Zhang Shouyi delivered a keynote speech titled "The Three Pillars of Artificial Intelligence". Zhang Shouyi said that China has a good opportunity in artificial intelligence because China has a huge field of data, mathematics and science. Many of the talented people, as well as China can use the outbreak of artificial intelligence to invest in the field of materials science and mathematical algorithms.

Zhang Shouyi believes that the explosion of artificial intelligence today is due to three important reasons. One is the exponential growth of computing power described by Moore's Law, the other is the massive data brought about by the explosive growth of the Internet and the Internet of Things, and the third is the rapidity of intelligent algorithms. development of.

Today's improvement in artificial intelligence algorithms, financial, education and other big data in a large number of production, but the data and the use of data are usually behind different companies, do not necessarily fully trust each other, resulting in a lot of data can not be analyzed in real time. Prof. Zhang Shouyi recommended a very novel algorithm—homomorphic encryption. This new algorithm can learn the wisdom inside the encrypted data. It does not necessarily need to see the data itself, making the owner of the data and the data processor complete. Can be separated and can establish cooperation based on trust.

When does machine intelligence surpass humanity? Prof. Zhang Shouyi said that this is misleading by the Turing test. The machine cannot and does not need to achieve complete imitation of the human brain. It is still a long process. If the machine learning part of the human rationality is easier, but the irrational people Emotion is not so easy.

Prof. Zhang Shouyi also set up a venture capital fund, Danhua Capital, to focus on the highest-level technology investment. He is determined to combine science and industry better. He believes that the highest aspiration of science is simple and universal. It is necessary to use first principles. Think about problems and invest. Really use the excellent education and scientific research of the school to make the artificial intelligence industry a good job. Prof. Zhang Shouyi said that he is willing to play a role in connecting the academic world with the corporate world, Silicon Valley and China.

The following is a record of Professor Zhang Shouyi's speech:

Thank you, everyone, and I'm very honored to share with you. The topic of my speech today is "The three pillars of artificial intelligence." In the past 100,000 years, we have been the most intelligent species on the earth. We do meet today. A new challenge is that artificial intelligence, the wisdom of machines, to a large extent, may exceed human wisdom.

Why in the past 100,000 years or so, at this time of today there has emerged such a new, magical species, mainly the convergence of three major trends, the first is the ability to calculate, since the computer was invented after the calculation of the computer In the past five or six decades, capacity has grown substantially in accordance with Moore's Law. The so-called Moore's Law means that our computing power has doubled every 18 months. This amazing ability to improve computing has enabled us to reach an era of rapid advances, and we can perform calculations that cannot be counted. This is the first trend in the convergence of three major artificial intelligence trends.

In the second trend, since the Internet has produced a large amount of data, artificial intelligence must learn, and it can be learned automatically through a large amount of data collection. The generation of data is due to the creation of the Internet and every corner of our lives. Become data.

The third trend, for me, is the most amazing development, some very new algorithms, people will initially simulate the principle of brain work to make some artificial intelligence, machine learning algorithms, but in the future, I today The topics that we mainly share with you, in the process of future development, we may introduce some algorithms that the brain can't achieve at all, but it can be implemented in the machine. For example, I talked about quantum computing and completely new algorithms in mathematics.

Our era is marked by a very great information theorem. The theorem is the Moore theorem. It means that computing power doubles every 18 months. Moore's theorem laid the foundation for the rapid development of the information society. In contrast, traditional industries are often In the more parallel, linear growth case, Moore's theorem refers to the information computing capacity growing at an exponential rate.

There are various phenomena now, and Moore's theorem continues to advance in the traditional way. It cannot be an exponential growth in accordance with the past development. Why? In the course of each transistor operation, a lot of heat will be generated. If the basic principle of the transistor is not changed, the number of transistors will double every 18 months, and the heat generated will double every 18 months. In this case, the entire chip will continue to rise and will inevitably burn. This is a very big challenge and crisis. I, as a physicist, saw this is a wonderful opportunity, so that we can completely return to a blank page and think about this principle. Why did the computer follow the original Moore? As the theorem continues to advance, what are the obstacles encountered?

In the simplest terms, we have built an information highway for electronics at the very bottom of the chip. The most fundamental principle of the highway, the principle of the lanes, their respective paths, and mutual interference, each electron is fixed. Running on the driveway, unlike the left, is a collision in a busy market.

Here I simply report to you what the scientific research invention I am doing is. In a simple word, the quantum spin Hall effect is to put the quantum on the chip level, according to the red and blue, the electrons themselves upwards, according to the red track chip in operation, spin down in accordance with the blue When the track meets impurities, the first lane is divided, and according to the spin, there is a spatial separation, but the direction of the spin is not the same, and the impurities continue to move forward and will not be scattered back. This is indeed a very recent discovery in the field of materials science and quantum science. We are looking for new materials, pre-researching new materials, and you may have heard of a magical material called graphene. It is a material made of honeycomb cells made of carbon atoms, but this graphene does not have us. The magical nature of the talk, but in the periodic table of elements, we will encounter the ene with the carbon atoms going down. This atom, like the carbon atoms, can also form a single atomic layer. We can preliminarily study this material as high speed. The highway runs, and the effect can be achieved at room temperature.

Now I share with you the artificial intelligence in the algorithm, I would like to first make an analogy with everyone on artificial intelligence. We think of the main reason for doing artificial intelligence today, and we want to imitate the brain of a person. We will look at the past of artificial intelligence and the future of artificial intelligence in the future. I want to explain it to you through an analogy because we see today. The human brain wants to imitate the human brain, but when we saw the bird flying 500 years ago, we wanted to learn. We wanted to ask a question, would people fly, fly like a bird, or could we make it? An aircraft. First of all, we see that nature's creatures have such a miraculous function. We ourselves think that we can come to bionic. But at the beginning, we learned to fly simply to imitate the flight of birds. We could later really make a magical plane. I heard that we will have dinner with the scientists of a big plane this evening. We are able to make planes today. It is not that we simply learn how to fly. We really want to understand what the mathematical principle of flight is. Because the mathematical principle is behind the so-called fluid mechanics. Fluid mechanics has a physics equation set. Once we understand it, After the mathematics of flight, we designed the aircraft to be able to fly better than the birds, but it doesn't have to fly exactly like a bird. We now design an aircraft that is completely like a bird but it is rather troublesome, but the design An aircraft that is faster and taller than urine is relatively easy.

The same thinking, to think of artificial intelligence today, the first step is to use neural networks to imitate every neuron in the human brain, or simply imitate the time, what should be done in the next step? What really understands the mathematical principles of intelligence and intelligence is like we understand the mathematical principles of flight. Once you understand it, you can design better algorithms. These algorithms are not necessarily what the brain can achieve, but they can do better than the brain.

If we want to clear this point, we can think of a real good algorithm to replace some of the technical bottlenecks. For example, in the field of artificial driving, we all know that most of the technical routes are still studying the technical route proposed by Google in the past. For example, artificial driving is equipped with a laser radar on the top of a car, which requires a high-definition three-dimensional map. It has brought a big bottleneck to the entire auto-driving field. Because first of all, Lidar is very expensive, and you have to create high-definition 3D maps is also very troublesome. Dan Hua Capital invested in a company Autox, which is able to recognize everything around it through ordinary cameras. Why can it be done? There is a simple reason, because people can drive, people are also made up of atoms, and robots are also made up of atoms. There is no essential difference between humans and robots. If people can drive, they don't need laser radar, and they don't need HD. If the computer can do the algorithm well enough, it can still be achieved. Now in Silicon Valley, Autox is a very good company. It is a Chinese professor, Professor Xiao Jianxiong, a doctoral student of MIT, a professor of Princeton. He has now left and founded such a company.

The emergence of the era of big data today has resulted in a large number of financial data, education data, and health data. However, the ability of machine learning nowadays is to learn the wisdom from this big data and help We have greatly improved our effectiveness in the financial, education, and health fields. But now I have encountered one of the biggest problems that can deal with data and can have data. These are two different groups of people, or two different companies. There is not necessarily a complete trust between each other, so that a lot of data cannot be analyzed in real time. In the field, I also recommend a very new algorithm, called homomorphic encryption. This new algorithm can learn the intelligence inside the encrypted data. It does not necessarily need to see the data itself. In this case, it makes the data. The owner and the data processor are completely able to separate and can establish cooperation based on trust.

Now, the question that everyone wants to ask in the field of artificial intelligence is whether the wisdom of a real machine will surpass people on a single day. How can we measure the machine over one day? For example, in the past two years, we have witnessed a very ambitious man-machine battle. Google’s DeepMind can defeat the highest human player in the next go. Does this already indicate that computers have surpassed humanity? One of the most classic test methods is the so-called Turing test. The Turing test is whether I'm behind a behind-the-scenes robot or a real person. I don't know, but after I talk to it for a while, can you find out if it is The machine is still human. If I can't figure it out, it means that the machine has reached human wisdom.

But I feel that this test is still somewhat misleading. The wisdom of the machine, if it can completely imitate people, is still a long process. The evolution of people in the past through millions of years as a biological species has indeed brought about some very rational ingredient. However, it is relatively easy to let the machine learn the rational part of human beings, but the irrational feelings of scholars may not be so easy. I think that the highest wisdom of mankind is that such a great scientist as Einstein could write so simple by simple formulas that we can verify by doing experiments. This is the highest crystallization of human wisdom. I asked whether the next time the robot could write a very profound theorem that was equivalent to the universe. As a scientific discovery, it was in front of people. In this case, the wisdom of the machine was truly more than human. In this field, we must make great development. We have clearly seen in the United States that we must make a very close combination between the academic and industrial circles because academics often have some smart talents who can come up with these The best algorithm, but now the industry, especially Internet companies, like NetEase, and China’s BAT, have a lot of data in their hands, and they must be able to do artificial intelligence even more during the process of close cooperation. Good next step.

As I thought of this, I began to take a step forward in my personal life. In addition to doing research and education besides teaching at Stanford University, I also founded a Danhua Capital, which specializes in doing The highest technology investment, so our most basic concept, first of all reflected in our name, Danhua, "Dan" is the meaning of Stanford, "China" is China, I want to establish the role of Stanford University and the bridge between China. Our basic idea can be reflected by LOGO. The triangle means the meaning of production, research, and research. We must truly use a school's excellent education and excellent scientific research to make the artificial intelligence industry a good job. We can also see that there are two cross-border areas, the two greatest cross-border areas of our time. One is to integrate science and investment closely. The other is to integrate China and the United States closely. Will definitely get a lot of opportunities. In this case, after the academic community and the business community have truly achieved a close integration, we can really advance the field of artificial intelligence.

Finally, I concluded that artificial intelligence is a very good opportunity for China. First, we have many Chinese people and we have very large data. The level of education in China is now international, especially in the field of mathematical science and technology. The international community has gone very forward. We have very talented people in higher education. I said that there are three pillars of artificial intelligence, one is big data, and China has very large data. One is the need to make big advances in physical materials. I also want to advance China's basic sciences greatly through the development of artificial intelligence. In addition, our algorithm requires excellent talents in mathematics. China is indeed a very good talent in this area. To achieve this, we must establish the role of two bridges. One is to connect the academic and business communities. The other is to link Silicon Valley and China together. I am willing to play a bridge role in this field.

On July 15, 2017 Netease Future Technology Summit was held in Beijing.

The theme of this year’s NetEase Future Technology Summit “New Life” is to point out that the Internet industry is recovering from the capital winter that has spread for two years. Large companies are accelerating reforms. Unicorns are emerging in an endless stream. Young entrepreneurs are catching up and reborn. The Internet has brought a whole new future.

This year's NetEase Future Science and Technology Summit set up six forums: New Technology, New Future, New Content, New Entertainment, New Consumption, AI+ Finance, AI+ Travel, AI+ Life, and AR Future. We invited the most outstanding scientists, entrepreneurs, investors and cross-border stars at home and abroad to discuss the future of artificial intelligence, consumer upgrading and AR.

No-encapsulated Sensor

No-encapsulated Sensor

No-encapsulated type NTC temperature sensor with the properties of sensitiveness, fast response, high temperature resistance, good appearance and protection has been used into oven, air conditioner, induction cooker, refrigerator and so on. Temperature range is from -30°C to 250°C.

No-Encapsulated Sensor,Oven Temperature Sensor,Air-Conditioner Sensor,Refrigerator Sensor

Feyvan Electronics Technology Co., Ltd. , https://www.fv-cable-assembly.com