Li Deii, academician of Chinese Academy of Engineering: Cognitive calculation in large data age

Absrtact: MDCC 2013 Mobile Developer Congress was held in November 2013 13-14th at the Beijing National Convention Center, which was co-sponsored by CSDN and innovation Workshop. More than hundreds of domestic and foreign mobile Internet leaders and core technical experts, to the national movement, heavy

The MDCC 2013 Mobile Developers Conference was held in November 2013 13-14th at the Beijing National Convention Center, co-sponsored by CSDN and innovation Workshop. Over hundreds of domestic and foreign mobile Internet leaders and core technical experts, "people move, reshape the world" as the theme, to share the latest practical experience, explore the development trend of mobile Internet. In addition, smart hardware, mobile entertainment games, micro-app three major theme summits to become the new highlight of the conference.

Li Deii, academician of the Chinese Academy of Engineering, published a speech titled "Cognitive Computing in the age of large data". He analyzed whether cognition could be calculated, and explained how cognitive computing should go in the big Data age and the mobile age. He believes that the era of human-centred cognitive networking has arrived, and cognitive computing in the big data age is approaching cognitive science.

　　

The following is a transcript of the speech:

Li Deii: Good Fellow. Mobile life, mobile quality, bring us a great change. This year people in society, get a mobile phone, laugh and laugh and have a photo, this is to bring us changes.

People live to move, the speed of the movement is how much, the mobile speed to make clear, mobile products to get clear. We rely on two legs to move, rely on the wheel to move, by high-speed rail, by aircraft movement, mainly or low-speed movement, so we call move, not called movement. But memorizing a notebook is a burden, so I think about what the big data age has brought to our mobile life, that is, the cognitive of people in the big data age, the simplest of which is speaking, listening, reading and writing.

Can cognition be calculated?

　　

This is a very old proposition. Because old scientists think people are the highest quality in the world, better than anything else, cognition is not computable.

We want to improve people's quality of life, we hope wisdom. To turn intelligent cities into smart cities, intelligence is a technology, and we want wisdom. We say that your conscience is greatly broken or I am not in a good mood to learn, which means that the cognition and mentality are related. However, as the science of artificial intelligence, as a cognitive science, we have long emphasized that IQ, brain intelligence, based on existing knowledge of analysis, association, deduction or induction, to produce information. We think that people's cognition depends mainly on the brain.

Later said no, you this person IQ is very high, eq too low not. During the interview, the boss should make intuitive and comprehensive thinking through imagery, intuition, imagination and emotion, which plays an important role in innovation. This is two yuan. More people say success equals 40% IQ and 60% eq interactions. So there's still a lot of cognitive science going on here.

In the 1936, the genius scientist Turing established the concept of Turing, any form of system can be accurately portrayed by Turing.

People usually give computing work to the computer, is it based on the intelligence of the human brain? IBM studied a production computer. People put people's thinking ability into the machine, the front desk is a machine, backstage is a person. In this sense, cognition should be computable.

Our program problems, downtime issues, and program validation issues are inherently not possible to calculate. So we patched and we had holes.

Mathematics itself is an inexhaustible machine program. Not all mathematical problems can be done by computer. So the human brain cannot be materialized into a computer, it is important whether the human brain can be formalized. Many Nobel laureates have made significant contributions in this respect. But the computational nature does not depend on the choice of the formal system.

Therefore, it workers must not use administrative methods to study thinking and awareness. How does consciousness understand? How does the soul understand? How does religion understand? How do you understand faith? How do they relate to thinking? There is a division of labor in the brain, this piece of what, what, then the results of the study found that people's consciousness and mental activity, is produced by the interaction of different regions of the brain. How to work together, the question mark is still very big.

Life scientists like physics, like chemistry, is the human mind governed by material and chemical laws? This is doubtful.

Some people think that it is impossible for a man to figure himself out. You have to study a thing, to stand outside one thing.

Challenges of the big data age

Under such circumstances, Nature magazine published an important article in which a small headline--bd (large data) The science of the Times, what we should do, this is the challenge of our time.

The source of large data is divided into three major chunks:

　　

the first block is the Earth, we call the digital Earth, natural large data to our Earth, the universe clear. The second is that life, including all kinds of animals, especially our human beings, is in itself a big data. The third is social, internet, especially mobile Internet, our daily life is big data, such big data makes you to the science, the Cognitive Science provides the challenge.

　　

We go to the doctor, now is not a doctor, the Doctor is very valuable, mainly to see the instrument. To the hospital has not sat for three minutes, the doctor a large pile of tests out, let X-ray, B-Ultrasound, endoscopic examination, nuclear magnetic, you have to queue up for two weeks, before the inspection done. You take the test results to find a doctor, the doctor less than three minutes to open the medicine, now Doctor is not a doctor, now is to check, the treatment of diagnosis. The Doctor tells you a word, you still do not understand the reception of the next patient, this is our daily doctor, pay to see the instrument, this is the reality of large data.

Over the years, many of the medical Nobel Prizes have been invented by medical firms, so today's smart doctors are either inventions of devices or they can discover value from medical data. You stand in line for 5 minutes and the doctor sees no problem, he finds value in this data. In the big Data age you'll find that you know what to do when you know what it is.

So the question is diluted, why not careful diagnosis, because so many medical records, you this situation in his patient library is very clear, so how to do is very simple.

We believe that science is the metaphysical solution, the technology is to solve the metaphysical, we study the object of cognition, wisdom. Now put our object is what the problem, can be solved with large data, in such cases, the large data on the different cases of the processing, can also use large data to solve. So the society has more to do with relevance and science in the back. Solve the causal problem scientifically, solve the primary and secondary problems and solve the problems before and after. So people can solve more problems is a hero.

We have a suspicion that science is the forerunner of technology. In today's situation, because of the development of IT technology, technology can also become the forerunner of science, demand the digitization of self-study, this is our large data on cognitive science, and cognitive computing differences. Cognitive computing is technology.

In this case, the large data age is leaving the cognitive science with computing.

In terms of human cognition, the European Union has studied brain science in the last century. This year Macao has proposed a human brain plan, and Obama has also launched a brain project. It's too little associated with our information workers, all life scientists, and I think our human brain, from the point of view of life science, is not much different from the monkeys ' brains, because studies have shown that the brain structure is similar.

I was thinking, I was studying in England, and Thatcher later had Alzheimer's disease, and I personally think that her dementia is not much different from the average person's Alzheimer's. Because I think the study of brain planning is a bit biased.

Compared with other animals, there are inspiration, language, only people have words, because there is the text, there is civilization, to inherit. It is important for our young people to learn knowledge and to learn civilization when they come to college. If there is no writing, we are about the same as other animals.

In addition to study the social attributes of human cognition, these two are not clear, cognitive science is not clear, I suggest that these two points as a point of human cognition.

To study the acquired attributes of brain cognition, cognition is not only determined by the nature of the preceding, but also by DNA, genetics and heredity, which depends on the result of learning and accumulating, which is related to social environment, social practice and group exchange.

What do we do in the age of mobile life?

Natural language processing technology in the big Data age is changing our mobile life.

Language is the direct reality of thought, this sentence is Marx said, language is the carrier of human thinking, language is the primary problem in cognitive science and cognitive computing to deal with. Our mobile life is first to speak language, you do not speak any day, can listen to say that we must have the ability. Let's face the advent of the voice age. Of the four major milestones in human civilization, language and words occupy an important place.

The effect that human beings use natural language to communicate. The effect that you communicate with others, it is important that your intensity and intonation, take up 38%, and your facial expression, it has emotion inside. How much do we pay attention to in the study of half a century of natural language understanding from our cognitive science?

When people think and solve any problems, they are usually qualitative, and the quantity of the regulation is often a non-numeric value. We study large data and it has a qualitative, quantitative conversion problem. In scientific research, people use object language to express a specific precise discipline, such as physical mathematics, C language, are expressed in natural language.

We become the argument of two schools of thought that natural language is not computable.

If a problem cannot be fully formalized, can local problems be formalized? The extent to which natural language can be formalized depends on the formalization of uncertainty. The study of uncertainty, I think, is a way.

Hkust News Fly made a customer service center, mobile customer service recording large data, can achieve a second to 1 million hours of audio data inspection. This is a big change. Are you hiring 500 customer service personnel or are you enabling a single server? So still use a server. How much attention have we paid to the study of half a century of natural language understanding?

We read the newspaper, there is such a passage, but you look closely, found that the word sequence is wrong, but it can affect us how many things? How much attention have we paid to the study of half a century of natural language understanding?

Another example, we are not always online to distinguish each other is a person, we often come out a recognition code, ask you to re-enter the identification code, prove you are human. In the study of half a century of language text understanding, how much have we studied this?

I think it is necessary to innovate and new ideas to solve. Robots can talk, they can think, they are changing our lives, this is the product of mobile life.

There is a product you can do, called online service robot. In the future IT industry, with the development of intelligent technology, mobile customer service center can be reduced to 50 people. Can we have a smart headset, this is my personal thinking, you wear headphones, I speak, my accent is heavier, your headphones hear Mandarin, you can do this? This is very good, especially in Cantonese, the North people do not understand, especially language translation. If the two heads of state talks, Obama's words are mandarin, but Obama's voice. That's a lot of change. In fact, there is already, this time need cloud computing, need a program service center, need a Chinese Translation Service center, Location Service center and so on.

Voice to text, text to text, we want to let the data speak, this is the big data to our mobile life brought about by the changes. So someone said to kill the text message first.

My age for five font input is a special feeling, we have two generations of people are using five font input. From search engine to personalized search, from rule learning to statistical learning, from intelligent computing to emotional computing, from formal linguistics to intelligent translation.

We imagine the future television host he will be able to read, the virtual host in the image, emotion, movement, voice, intonation, style is like a certain person how good, then we actors can also be unemployed. Such an era may come.

The language which is reflected by the large data form is an original ecological language which can be interacted and understood, which is hairy, vivid and emotional.

Intelligent driving

Let's talk about the intelligent driving in our mobile life, why does intelligent driving fire up? Because it is closely related to cloud computing, mobile Internet, big data, you just do this thing, the government will certainly support you. Because it is a strategic emerging industry. When the Natural Science Foundation in determining this project, proposed the scientific task, solves the actual computation question. Later put this scientific problem explicitly with a carrier to performance, called scientific tasks, under normal circumstances, Beijing to open to Tianjin, you can do it?

We used a lot of radar, we have to have a number of driving behavior, in this process we encountered a lot of difficulties, such as to say to develop a three-dimensional imaging, how to have self-learning function. We have achieved gratifying results.

We have detailed experimental reports, experimental environment, and the number of overtaking, we also played with the Korean car this year, we made an interactive identification in the car.

Let's think about it. Mobile phones are now for human use, so later it is human and machine interaction, in the future, you use mobile phone, you put your cell phone into the car, the car ran away, to a place, the money to pay you can go, you change the path, you use mobile phone and robot dialogue, we are here entrepreneurs can be made, people, The relationship between mobile phones and wheeled robots. If this is done, it will be a killer.

My idea now is to do these two things, Beijing from Tiananmen Square to the capital airport to do an experiment, mixed in the normal path, at present we have 5 cars. Shanghai We hope to go from Hongqiao Airport to Pudong to the people's Mall, we have 6 cars ready. One is the electric vehicle, one is the hybrid car, the other is the domestic car. One type two vehicles. If one day North Beijing appeared a motorcycle racing robot, of course, this is illegal, you have to ask cognition can calculate?

In the age of large data, the effectiveness of technology is more important than the integrity of science. So I suggest this picture (see ppt below), to some small and medium-sized enterprises Some reference, I think people-oriented cognitive networking era has come. There is a smart glasses to solve the camera problem, to solve the problem, there is a smart headset to solve the problem of listening, there is a smart pen to solve the problem of writing, as well as smart watches, to solve your physical testing, as well as the needs of communication and so on. In this way, people-oriented, forming wearable equipment, so that can do a lot of things. This really can improve the quality of our mobile life, the realization of intelligent life.

To sum up, it is a matter of cognitive science to say that human beings can not fully and uniformly formalize cognitive problems. Our cognitive computing allows for the local formalization of thousands of small and even thousands of specific situations. So is cognitive computing in the big data age approaching cognitive science?