Big data and artificial intelligence development thinking

Source：Updated：2017-10-20 08:39:55

On October 12, the 7th China intelligence industry peak BBS in foshan, in the first day of the Lord on the BBS, Beijing TRS Information Technology Co., Ltd., vice chairman and President of water just published a theme for the big data and the development of artificial intelligence thinking's excellent speech.
Big data and artificial intelligence development thinking
In his speech, Mr Shi water from their years of big data technology and service leadership point of view, introduces the launch of its use of big data technology data value-added services platform, and it is concluded that the "data, information, knowledge, intelligence, wisdom" the value of the path of ascension. From big data and cloud services to artificial intelligence, it is of great reference value to provide a way for guests to use big data to realize the value-added of artificial intelligence.
 
The following is the presentation of Mr. Shi's speech:
 
Good morning. Thank you very much for inviting me to share my report at the conference. Today I'd like to share with you some thoughts on the development of big data and artificial intelligence.
 
The first thing I want to talk about is that I think it makes sense to compare the big data to the artificial intelligence industry and to put these two things together. The second point, to speak of what we are artificial intelligence + industry, or industry + artificial intelligence, talk about my understanding and awareness, the third point I think we need to break through artificial intelligence now very emphasis on three elements, is computing power, data and algorithm, I think for the future research and application of artificial intelligence, only these three points are not enough, there should be other important factors need to be added. Fourth, I want to discuss is that we now in several directions of artificial intelligence, which have big opportunities, to make our innovation, entrepreneurship, make money, and finally something about our own some artificial intelligence based on NLP platform application practice.
 
The contrast between big data and artificial intelligence
 
It is instructive to compare the development of big data and artificial intelligence industry. Because the development of artificial intelligence and Data are inseparable, and now most of the achievements of artificial intelligence development and is closely related to the big Data, thus observe the development of artificial intelligence in the big Data industry industry development is very meaningful, at the same time, we think the data-driven Business (Data Driven Business) than intelligent drive Business more in line with the nature of industry, big Data industry actually landing ability is stronger than the artificial intelligence, so big Data problems arising from the industrial development of artificial intelligence industry development is very meaningful.
 
The development of big data has several aspects to the development of artificial intelligence. Including the importance of data, importance of data quality, importance of application scenarios, importance of industry knowledge, importance of policy regulations, and the reference meaning of the model of liquidity. Big data from the White House in 2010 in the United States first began to make some policy, our country began to heat up to 2012, over the years introduced a lot of policy specification, even make a lot of park, but we found that the whole big data industry now is still at a very early stage. Why do you say that? First, in what ways does it promote industrial change? Second, who made the money? Now basically only the big Internet companies recommended by precision marketing, e-commerce, and so on to make money, but we are a large number of engaged in the industry enterprise most large data in burning money, industry is also not benefit from the large data and is a major industry change, still at a very early stage. The same is true of artificial intelligence, where most AI companies are still investing and burning money.
 
Let's take a look at four factors that affect the development of the whole big data industry: data opening, technology r&d, industrial ecology, and laws and regulations. Generally speaking, the current big data very early industrial development, is still a big data investment and business opportunities, on the industrial ecology is the main characteristics of monopoly and new data islands, big data startup still need 3 to 5 years to achieve scale profit, most continued to burn, 2017-2018, industry consolidation trend is obvious. In terms of data, Internet data of big companies, hegemonism, government data open difficult (60 is very backward, and in the international away), industry and enterprise data is difficult to obtain, and the grey data gray industrial chain, and personal privacy are very outstanding, our country each year data trading market is more than $500, but legal only around 10%, 90% are grey data link, so led to the current public security check, said a lot of big companies was arrested, personal privacy is very outstanding, to decode the data of the curse, need in the laws, regulations and industry ecology two perspective.
 
In addition to the data and the quality of the data, application scenario is very important, is not important, the data of four V Hadoop/Spark is not important, important is application scenarios, so it's the same for AI, because we actually see the popular applications of big data and AI actually high contact ratio: finance, health care, education, online advertising, intelligence analysis... Application scenario is closely related to industry, are mainly vertical and from where, once in the industry, you will find that a lot of problems come, so we say that artificial intelligence development only emphasize data, calculate the force, the algorithm is not enough.
 
Policy law is also very important. In the Internet age, why is the Internet development in our country fast? One of them is a lot of people, the demographic dividend, and a very important rule is not so strict, the government and industry have more support for the Internet, and sometimes even a little contempt for the rules at the corporate level. But big data and the era of artificial intelligence, like the early days of the Internet, have not worked. Now data is open and privacy is protected, from privacy to personal safety, because it can kill people.
 
For a second, by comparing the development of big data and artificial intelligence industry, we can get a few conclusions: the artificial intelligence industry is still in a very early stage; The importance of data is undeniable, but there are many problems. The application is the driving force; The vertical sector is where most of the players are.
 
Artificial intelligence + industry or industry + artificial intelligence
 
The second thing I want to share with you today is about whether it is "artificial intelligence + industry" or "industry + artificial intelligence". My basic view is that industry + artificial intelligence is still the mainstream of the development of smart industry. We think that "industry + ai" may account for 90%, and "artificial intelligence + industry" may account for only 10%. What's the difference? The artificial intelligence + industry is creating new models, more of which are not mature in the previous industry, or have no existing good business models, such as autonomous driving; And the industry + ai is to transform and transform the industry with artificial intelligence technology, or reduce the cost, or improve the decision-making and management level. Like law, education, finance. Very of AI in the field of new technologies in the field of consumer and industrial applications, such as face recognition, facial cameras in the field of consumption in the industrial field may be identity authentication, bank accounts, security monitoring, analysis of business; Voice recognition and input technology in the consumer domain may be intelligent customer service applications, and deep learning and image recognition can be used in the industrial field as smart sorting and security screening.
 
If 90% of opportunity lies in the "industry + artificial intelligence", so money is not decisive, data and algorithm, if the money is the decisive factor, it also won't have any innovation in the future. Industry knowledge and industry experts are barriers, so the startup companies under the oppression of big companies still have a wide space, so those industries will achieve the AI outbreak or the biggest impact? Member of de-yi li said just now is very good, the four industries, manufacturing, education, finance, health care, I very agree with, I think the key is to look at two points, one is the growth of the industry itself, that is big enough, whether the future growth, the other one is whether the industry depends on the person's experience and knowledge, rely on, the greater the replaced by artificial intelligence, the greater the demand, such as doctors, mostly experts too. We invested in a project, it is pathological cancer diagnosis, it is said that the country can see the doctor the doctor is less than 10000, compared with less than 500 qualified experts, you see how much demand, big data and artificial intelligence can solve this problem, IBM Watson's idea is the same. The other is a lawyer and financial industry, the main is too expensive, a lawyer for an hour how many money, often to millions of financial industry, in fact you with the big data and artificial intelligence, they is not worth so much money. Academician li said just now far lagging behind in the number of manufacturing robot used in our country such as Japan, South Korea, the main reason I think it's cost, so I think the most pressing may not be a domestic robot, but instead of the high cost of human and the human is not enough.
 
The ai industry also has a distinctive feature. It's Embedded and Embedded, so AI Technology is a kind of Enabling Technology. In the future, All enterprises should be AI enterprises, so you can see that now including GuGe baidu, they say they are AI All In.
 
Three elements of artificial intelligence
 
The third point that I want to share with you today is the three elements of ai. Now, artificial intelligence, like the four V of big data, almost every expert will have to use the three elements of artificial intelligence, data, computing power, and algorithms. But is it enough? What I think is not enough, why say, look, and artificial intelligence are perceiving intelligence to cognitive intelligence from a computational intelligence, and create intelligent direction development, including the ability to understand and use the language cognitive intelligence, knowledge and ability to apply the knowledge, in language and knowledge reasoning ability, focus on linguistic intelligence is NLP. When it comes to cognitive intelligence, there are data, algorithms, and computational resources that I think are not enough. What do you need? I think it's important that you need a lot of knowledge, you need to have a knowledge map and so on, so knowledge is probably the fourth element. From another Angle, at present the three elements of artificial intelligence in not discuss application scene, and the basic or from the technical level to consider the problem, we think that is not enough, so the problem of application scenario could become the fourth element? In addition, people must consider human problems, artificial and intelligent, and the problem of human-computer interaction must be considered in many AI scenarios. So I ask a question, how do you look for the fourth element of artificial intelligence?
 
The key opportunities and direction of artificial intelligence
 
The fourth point that I want to share with you today is the focus and direction of ai. This is cognitive intelligence represented by natural language processing, or NLP. The AI in the field of hardware investment, is also a big company in the world, such as the GPU, FPGA and ASIC chip, such as investment in the field of vision, also many, especially in the image recognition, speech recognition, a lot of the birth of the unicorn company, technological progress quickly, but too much. Not the future of investment. The cognitive intelligence, which is based on the natural language processing, has a large gap in domestic investment and development. According to tencent institute report, the United States in the field of NLP DE novo actually is almost three times higher than China's, according to wuzhen think-tank report, 2000-2016, the global accumulated new data of natural language processing enterprises reached 543. Every year since 2009, the natural language processing enterprises accounted for the proportion of the global new companies remain at around 40%, according to research firm CB Insights recently released "the most notable 100 artificial intelligence companies, about 25% of the projects and" natural language processing "directly or indirectly related, so NLP has become the most notable artificial intelligence company. According to Forbes' choice of Top50 AI, which raised $34.15 billion in 2016, about 16.2 percent of the funds were directed to "natural language processing" directly or indirectly.
 
Why is that? Because in the AI field, relying on large data and deep learning, progress is the fastest, the best effect is identified, the machine machine identification combined with machine learning in many AI application scenarios to achieve the better machine intelligence, but the recognition is not equal to understand, in the future to achieve machine intelligence or "wisdom", also need to solve the problem of understanding and reasoning machine, especially natural language understanding/generated (natural language processing technology). How to let the machine like a human thinking, to understand human language, using human language to express, and emotion perception, reasoning, planning, decision-making, has the capability of self-learning evolution these are the difficult problem of NLP.
 
NLP contents is Paul from lexical, syntactic, semantic and discourse and language basic technology, the classification, clustering, sentiment analysis, knowledge map, machine translation, automatic question answering, information extraction, automatic summarization and other core technology, to intelligent search engine, customer service, public opinion supervision and so on NLP +, and then to the vertical line "NLP + industry". Early you think deep learning contribution to NLP is not big, a breakthrough over the past two years, we test on 6 data sets show that deep learning contribution to the automatic classification of the spring, the average can be increased by 5%, so the depth of the study is useful to NLP, but the increase of precision is not enough.
 
There are a few things that I think are very hot right now in the natural language processing, and the first one is the chat bot or the virtual assistant. There's a special BBS for tomorrow, especially good, I'm not going to talk about it here. Why do you say that? Because it is all the crest of natural language processing technology, and it is a focus of the future for the entrance, so apple, Microsoft, facebook and amazon are doing, and in the future this platform to open source.
 
In general, the current investment in natural language processing in China lags far behind investment in visual processing, which is an opportunity for us. But natural language has a greater barrier to dealing with barriers and a more cohesive industry.
 
Finally share with you about our company based on big data + NLP technology do some of the things, TaErSi is a big data + artificial intelligence for the development strategy of listed companies, the gem TaErSi artificial intelligence development strategy, the first is AI All routes, in which All of our technology products, applications and cloud services are embedded in the AI technology; The second is big data driven, which USES the technology and application basis of big data to support and drive the development of AI technology and application. The third is the application scenario, because there is no use for the technology without the business scenarios, which requires the application scenarios to lead and value the cash. At present, the scene of the main development of the technology of topology AI is in the vertical industries of finance, security, media, justice and intelligence. Moreover, TaErSi technology development path again emphasize two aspects, the first is the cloud service, gradually make business is a cloud service mode, the second is committed to vertical integration, if you can't form a vertical integration of ecological closed loop, you can't make a lot of money, to achieve a higher threshold for profitability and competition. At the same time, we also emphasized that we should use open source and open source frameworks and platforms as well as the achievements of our own specialized fields and innovation.
 
In conclusion, I think that the development of the big data industry is very meaningful to the development of artificial intelligence. Second, we believe that industry + artificial intelligence is still the mainstream