Archive for August 2020
Ali Cloud intelligent ecological partner MVP series logic
By Fang Jianyong
At present, cloud platforms include not only Ali Cloud, but also Tencent Cloud and Huawei Cloud. In addition to these universal cloud platforms, there are also some dedicated cloud platforms. A dedicated cloud platform may be applied in a specific direction, such as industrial use, military use, etc., with higher requirements for privacy and security. However, after the platform is built, how to effectively obtain platform users and generate great value, it is necessary to construct the ecology around the platform.
For example, smart city led by Aliyun, Urban Brain has invested a lot of research and development power in security and transportation, and the results of research and development have also been applied to some major cities in China. However, such self-application research and development model has several disadvantages for Alibaba. First, after the platform is built, it needs to employ a large number of talents to implement its own ecology or solutions built around the platform. First, solutions for different industries, such as urban brain, industrial brain, agricultural brain, etc., are also too heavy for Aliyun. First, talent needs to be sought and cultivated. There is no such thing as finding so many people to create so many applications all at once. Second, once employed, the cost of human labor in captivity is very high, and the people employed may not have a better understanding of the industry than outside professional firms.
There is a problem with solutions targeted at a specific industry. Internally developed solutions may not be as good as externally developed solutions. It is neither feasible nor necessary to use Alibaba’s own resources to compete with the national application solution resources. Of course, Ali also has some strengths in security and transportation mentioned just now, mainly based on the consideration of revenue, because security and transportation are mainly engaged in government projects, and the revenue accounts for a big part in the application of Aliyun platform, plus its gross profit margin and receivables can be guaranteed.
Security and transportation are relatively high-quality solutions for Aliyun, which it can do by itself. Security and transportation application solutions, with a high degree of concentration and little difference, have productization properties, are universal industrial application solutions, and Aliyun is also willing to do it itself. However, for other industries with more categories and deeper levels, Aliyun has no ability to construct, and there is no revenue worth expecting. In this way, the market segmentation industry solutions companies can participate in, which is more effective, Ali cloud to do its own core solutions, but also to build cloud computing resources, cloud storage, as well as a large number of algorithm model base platform, and operate well.
Ali Cloud intelligent algorithm model library is mainly in the charge of Dharma Institute, which will construct a large number of algorithm models for different implementation application solutions, including deep learning and machine learning. It concentrates open source models that can be rapidly trained and formed in the market at present, and encapsulates API applications constructed on cloud platform. These applications may achieve some good agile results at the moment, but as the business of applying solutions deepens, significant improvements in algorithmic capabilities are required, and it requires constant iteration. Algorithm iteration of work, may be ali dharma hospital to do in the future, but more likely to ali cloud intelligent ecological partners to do, because these models of the accuracy and optimization, it needs the actual application process of constantly to adjust, it will collect more data, and then by the business data to iterative algorithm model, coupled with the application scenario experience accumulation, model and application layer solution of software and hardware integration, it will be more and more deep, this is a dynamic development process.
Ali Cloud intelligence is equivalent to the integration of cloud platform and AI intelligence, through the construction of ecological MVP such a group, to realize ali cloud intelligence value-added. In the future, AIOT is a long-term development direction, and user data and solutions will have a vigorous development. Under the joint promotion of cloud intelligent platform and ecological structure, it is possible to draw close to several enterprises in the head or even one enterprise in the head in the future. The plan of Aliyun is more about this trend, that is, the development direction of the enterprise head trend. It is an idea of adding ecological structure as a platform and a logic of business development.
In addition, for ecological partners around ali cloud intelligence, also is the MVP, ali cloud intelligence without endorsement of ali cloud, he put up the fusion algorithm and application solutions to achieve a better performance, for solutions company, it is more difficult, because it will give the solution to do on the one hand, to be born, on the other hand, he also from a technical level to two aspects of cloud platforms and AI intelligent also well, very long enough. For the platform, AI intelligence is to build a relatively common thing, which involves human resources, material resources, financial resources, energy and other factors. However, for application solution companies, common performance is not required, but the industry segmentation and verticals are emphasized. Through cooperation with ali cloud intelligent ecological, solutions can be to put a lot of energy solutions, give the things to do, such as do, in the field of judicial solution in the field of justice is in place, industry, industry application solutions is in place, the agriculture, to put the solution of agriculture. We should focus on the core direction we need to work on to achieve a mutually beneficial outcome.
These are the logical interpretation of Ali Cloud Intelligence’s construction of MVP group of its ecological cooperation partners.
Analysis of application scenarios of intelligent garbage classification
By Fang Jianyong
The application scenario of intelligent waste classification is that one is household waste and the other is medical waste. Our first class living waste, this is because the government is pushing the living garbage classification, garbage to kinds of storage and transportation, in accordance with the relevant provisions and problems arise, and to comply with such garbage classification, we are on new equipment, new system, a new solution, or in a traditional, existing solutions based on an iterative update?
Let’s talk about the new one first, because it is a piece of white paper, which is relatively easy to make. The new intelligent household garbage classification has two ends, namely the front-end garbage can and the back-end centralized garbage treatment center. For us ordinary people, the bins that are closest to our lives are in the communities around us, or in public places. These bins have changed in many forms over the years. To comply with the new garbage classification regulations, besides having a distinguishable sign, there is also a designated garbage can combination. Through community education, we take the initiative to comply with the operation of the new garbage classification regulations. The other is the combination of human and artificial intelligence garbage classification.
If I am not sure what kind of garbage an item belongs to, Then I put it in the garbage release window, put it in the window will perform automatic visual recognition, and then prompt you to throw it into the designated trash can, this is a kind of. In the other case, you put the trash directly into one of the entrances of the trash can, and then it does a visual recognition, and based on the result of the recognition, it automatically sends the trash to the designated trash can.
The centralized treatment of household garbage can be divided into two types, one is meal waste and the other is non-meal waste. Food waste is relatively miscellaneous, not easy to be sorted, and difficult to deal with. The dirtiest and most onerous work in waste treatment plants or waste disposal stations falls into this category. We first have to deal with kitchen waste. , of course, in the front of the garbage classification, it has been confirmed the food waste, is no need for this type of waste itself further classification, existing rubbish collection and waste places, we can a equipment renovation, to a biodegradable kitchen waste processing, to the garbage recycling, generating clean biomass. The transformation has two effects: one is cleaner, turning waste into treasure; the other is that it doesn’t need to be concentrated in the designated garbage dump anymore, because it can be solved in the garbage transfer station, which not only shorens the garbage life cycle process, but also saves manpower and transportation capacity.
For non-meal waste, it is divided into toxic waste, recyclable waste and other waste. Garbage if it is through the front end, and as the toxic waste, recycling garbage and other categories of garbage sorting good, recyclable garbage on the conveyor belt, through the conveyor belt when visual identification, identify belong to some kind of spam more detail, such as recycling waste plastic bottles, mechanical arm will take the bottle up, thrown into a corresponding recycled plastic bucket and other category as well.
Another is medical waste, medical waste classification of existing actually have formed a closed loop, the hospital internal hospital logistics department has its own recycling processing functions of such a position, however, it is the final processing go somewhere for centralized incineration, mainly considering the exposure of medical waste, adopted the method of the save Labour when the province of one size fits all.
When the hospital collects medical waste, each department has a special medical trash can. Medical waste is basically thrown into a bucket and then incinerated. In this process, we put forward a concept of medical recyclable waste, which means that all medical waste should be recycled and used with low safety level medical waste instead of being incinerated in a simple and crude way. For example, recyclable medical waste such as bottles of brine, bags of brine, gauze, needles and so on can be distinguished from non-recyclable medical waste by garbage can in the front end. Therefore, we need to put different garbage cans in each department of the hospital, which can be recycled and non-recyclable.
For medical recyclables, we will implement a new process for recyclables, that is, there are two parallel medical waste disposal paths in hospitals. Unified to the centralized processing center, medical recyclable garbage into the conveyor belt, through visual identification and mechanical arm, the recyclable garbage sorting into a single classification dustbin, such as salt water bottles on glass recycling bin, saline on recycled plastic trash can, finally the visual recognition effect is poorer, or is unable to visual identity, or mechanical arm is not convenient to take, does not have the robotic arm to be able to sort, directly through the transfer to comprehensive health care to the trash can, finally to return to the original do not recycle process, finally to do centralized burning treatment, That’s the idea.
The promotion effect of Shared trucks on coal transportation under the policy framework of “carless transportation”
By Fang Jianyong
One of today’s topics is about some of the advantages of Shared trucks in commodity transport under the “car-free” policy. The bulk commodity transport in question is mainly coal, with the focus on Inner Mongolia, China’s main coal producing region, Shanxi Province, and the outer Mongolian frontier bordering Inner Mongolia. The main USES of coal are power generation and coal chemical industry. In addition to water transport, the main transport on land from the railway and road, today we mainly talk about road transport, talk about the use of coal chemical industry.
Coal chemical industry is a new application direction of coal application in recent ten years. Along the roads from Ordos in Inner Mongolia, northern Shanxi and outer Mongolia to Hongyazi, a coal-chemical base in eastern Ningxia province, a large number of small, scattered truckers look for supplies and look out for information from producers in need of coal. So what’s going to happen? The main reason is that the information is not equal and not timely enough. Since the ministry of Transport issued the “carless transport” document at the end of 2016, enterprises in the society now have the opportunity to do “carless transport”. I call it “Shared trucks” here, which means they don’t own a car, but use social vehicles to form their own capacity to transport bulk commodities and coal.
The good times in the coal business are over, and if you want to be a relatively profitable coal business, you first need to make it big, in sufficient quantities, and with sufficient efficiency to do it. We have so a “car free carrier” under the framework of sharing platform truck, with tens of thousands or even tens of truck driver register, platform and the truck driver can WeChat, APP, mobile phone, telephone, SMS, such as real-time interactive tools, let the truck drivers first learned that transportation orders, as the shipper can bring order to monitor the whole process of transportation, finally realizes the online paragraphs and playing in a timely manner.
For the Shared truck platform operation under the framework of “carless transportation”, the business type has evolved from the coal trade in the early days, opening up the whole supply chain of coal trade. Now, through this platform, customers and both ends of the coal mine can ask for better resources.
For example, on the customer’s side, under normal circumstances, more than 10 million level of deposit is required (coal power plants have this requirement, coal chemical industry and coal transport business here are also effective for them). Now, with the overall transport resources and complete supply chain capacity, we can talk to the customer to avoid the deposit. For the coal mine side, after accumulating a large amount through long-term procurement, ask the coal mine for a better price, 10 yuan or more per ton cheaper. In this way, under the circumstance that coal transportation and coal trade are relatively difficult to do, and the business is not so stable, we have developed a self-owned and efficient means which is in line with the current commodity transport, and we can achieve the volume of coal business in the early stage that can be predicted, reaching the level of trading volume of over several hundred million.
This gave existing coal traders a breakthrough direction. On under the condition of limited funding, the business is not stable, is to use the Ministry of Transport and state some of the new policy, and now owns the Internet, the Internet of things, artificial intelligence and other advanced technology, to develop a set of suitable for your company’s new business model of the traditional business, thus accumulation quantity advantage and long tail advantage, this is a new breakthrough, we also in improving or enhancing traditional business process of essential ability to improve itself, is also a big leap in its own core competitiveness. In this transformation process, the integration of traditional business and new technology can be formed to achieve a business type or business model update.
Released on 2020-01-03
About the development of children’s programming interests
By Fang Jianyong
I used to think that children’s programming was relatively simple, such as graphical and modular programming in Scratch, which could be taught by themselves. However, in the process of teaching children to learn Scratch once, I found that some concepts of variables and loops were involved, in fact, the children could not understand this thing thoroughly. At the same time, graphical programming, as opposed to code programming, is very different, and we haven’t taught it. Such oneself teach a child to learn in the home, the effect is not particularly good, the child also can say I actually also not how.
In fact, there are blogs and tutorials on the web that you can follow to recreate. In addition, you can visit the Scratch website, where you can download some scripts.
Scratch3.0 was released in 2019 with a slightly different but friendlier interface than 2.0, which was released in 2013, almost six years ago. We went to visit the MIT Media Lab, the lab that was led by Scratch inventor Mitchel Resnick, and it was a bit slow to get around the country. Especially the 50-megabyte Scratch desktop executable, which took hours to download. Fortunately, there is a domestic mirror on the Internet, the download is faster.
The Scratch website, which is also a cloud platform, allows for editable execution, as well as the ability to store finished scripts on the cloud platform for some sharing. The Scratch website has a number of works that have been classified and can be modified, saved locally, or saved to your own Scratch account on a cloud platform.
I thought, what should I do? It’s important for kids to get started in Scratch, how do we get kids interested? Kids have an interest in Scratch programming itself, but there are points of interest that we parents don’t know about, and kids don’t know about. Neither parent nor child can find a point of entry, making it difficult to spark a Shared learning experience.
Or need to have a get started so a person, and professional, Ningbo has a children programming training institutions, is the local has 5 entity classroom chain training institutions. When it was double 12, there was an activity, which allowed two consecutive auditions for one and a half hours each time, so I planned to give them the entry-level job. After all, at this age, teachers’ words are more effective than parents’ words.
In these two audition classes, the child showed a very positive behavior, and the teacher would also send a positive signal to the parents, saying that the child liked and was capable of accepting this thing. Judging from the performance of peers or beginners, it was still quite outstanding, which inspired the parents to sign up for the class. Through two free tutorials demonstrating Scratch cases, the child gained certain understanding and hands-on skills. When I got back home, I downloaded some Scratch scripts and went with him to talk about what they had learned in class, and went over what I had learned in the audition at home to consolidate.
A child will try to do all the characters, backgrounds, controls, sounds, loops, and so on in Scratch based on tutorials and videos. In the process of doing or trying, I will interact with him. In this process, I also know some ideas of the child, and I can get timely help if he does not understand. For really attract children in some places, I’ll find something related, such as books, Scracth 2.0 some more books on the market, the individual has been updated to version 3.0, of course, with the latest 3.0 synchronous Scratch, find a few of the domestic, also find some foreign translation books, good buy one is enough, shall we go to ningbo the library for the books more, after all programming books are copying once or twice, it doesn’t buy more cost-effective.
B has a lot of scratch-related videos on his site, and kids can practice them one by one, which is a very important part, because kids are more receptive to videos than pictures. During the rehearsal, the child will change some materials and express his ideas through this work. He will express this idea, in a sense, he is no longer a programming concept, but a practical means for a child to express an idea.
Friday, Saturday, or Sunday night to spend an hour, let him to understand, because the process of doing is actually quite time consuming, sometimes before you know it is one hour, may be two hours have passed, so it is time to control good, if has been done, let the child keep good script first, two times to do it.
The child to operate a few links, the importance may not be very clear, so want timely save, life good name, oneself build a folder. And in this process, it’s really about parents working with their kids, and also having the kids control their own computers and do some of the Scratch related things that they can do, and then maybe having the ability to do their own demo, something like that. Then we will be in training institutions 15 lessons, once a week, in the process of children is not only a simple repeat the content of the teacher speak, he would have to interact with the teacher, because before the start of the next start class in the middle of a month or two it just has a buffer period, during the period of our parents and children to Scratch children’s programming has a deeper understanding and practice, so that the rest of the class, the child’s learning effect will be better and more interactive.
Discrete crowdsourcing in the software industry
By Fang Jianyong
Today’s topic is a discrete crowdsourcing problem in the software industry. As for such a topic, I have been thinking about it for quite a long time and I have been confused for quite a long time. What is the main thing? Major changes have taken place in the software industry over the past 20 years, mainly due to the huge increase in human cost. The human cost of software development is likely to increase several times by 2020 compared with 2000. The model of a traditional software company is that it will take orders for the software engineers it hires to do, and the software companies can be thousands of people, or tens of thousands of people, or hundreds of people, or dozens of people, or even a few people. So if you have a software company of different sizes, it’s going to target projects of different sizes, and it’s going to take on different projects at the same time, right? It’s a tradition.
In the early development of the software industry, such as in Silicon Valley, Boston and Seattle, software developers, as knowledge-intensive industries, are relatively physically concentrated. With the development of the Internet and higher education, software companies have lower and lower requirements for distance. A software project in the United States, it will be done in China, it will be done in India, because the labor costs in China and India are much lower than in the United States. In the software industry, a more conventional method of workload calculation, or development cost calculation, is based on person-month, so in the software industry, its main cost is people and time.
The Internet solves a distance problem in software development, which makes the labor transferred to developing countries with low labor cost. This is called offshore software outsourcing. But the way software is distributed is still the same, it’s still a way of hiring, which is to say, in the developing world, a software company contracts for an outsourcing business from the United States, and it’s still going to hire software engineers full time, just like traditional software companies do.
So what is the current state of the software industry? China and India are the two important software outsourcing offshore development center, the labor cost of it is also growing sharply, especially in the coastal areas of China, especially north shenzhen first-tier cities, the cost of programmers may have been close to silicon valley in the United States, of course, at present there are some distance away from the silicon valley wages or, however, it is also in slowly close to that way. The problem is that the profitability of Chinese companies, which used to come in the form of software outsourcing, is also a relatively low margin. As labor costs soar, such profits will soon be wiped out or even lost, meaning it will no longer be profitable enough to undertake outsourcing from the United States. How to do? In this case, the impact is revolutionary, not only that, but also in some domestic software outsourcing business, in this case, the client is already in mainland China, it also has no profit and can’t afford to hire people. What to do? And as the number of programmers continues to increase, many software engineers may no longer be able to find an employable unit for their target, so they lose their jobs!
This is where the long tail discrete crowdsourcing effect comes in. That is to say, it can be a part-time job as a software development engineer, or its main business is not software development, or software development. On the other hand, I’m not employed by a software company, I’m a free software practitioner, right? That two people will exist in large quantities, software companies to cut costs, it will hire some of the more important role, such as product manager, project manager, the core of the architect, one can solve the problem of core programmers, to large, general software developers, it USES a form of temporary employment, in fact, is not affected by those people it employs, just when it needs to be timely and effectively organize the personnel, put these people don’t need to release. The cost it provides is nothing more than the amount of work between the start and end of employment.
This has several advantages. If I’m going to do a project, if it’s a traditional software company, he’s going to hire an equivalent amount of people to deal with the project itself, which makes it very difficult for him to find all the right people in a short time, right? No one to hire in a short time! If it lasts for the duration of the project, which is probably the life cycle of the project, after the project is over, if these people are hired, then they have to find a new business to support them, and that becomes a liability. This is a problem, and as I said, we need to find the right software engineers right away to take on different roles and deliver the project, right? So under this situation, we need to have a platform to match our demand, with our developer divide, the demand may be a big party, may also be a small party, the meaning of party a, which means it might have big contract down after party a’s business, and let the project by discrete programmers to an outsourcing cooperation. There will be some profit for them, won’t there?
So where does such a platform come from? Is there such a platform at present? At present, there is none in China. But there may be a better discrete crowdsourcing community for software development in the US called Stackoverflow. The community has survived for many years, and the users above are basically software developers. The personnel are distributed all over the world, and they can do some difficult questions on them, which is not only timely, but also accurate. In the long run, it is a huge knowledge base. In such a large platform, in enough time, different types of programmers have produced a large number of questions and answers, that is, the programmers who are the users of the platform, all traces of them are on it. Through the data mining of this platform, we can judge the performance of a programmer in the past history. The first is the programmer’s professionalism, the second is the programmer’s enthusiasm, as well as his activity, accuracy and so on. The platform will evaluate the programmer’s ability, and then match the corresponding project. On the right side of the web page, there’s a section called “Remote Recruitment” that describes how Remote employees are hired.
In addition to getting the right developers together, we need to monitor the development process effectively. Managing each of the roles in the project, monitoring their progress and continuously presenting the overall progress of the project, such as Teambition Team collaboration software, recently acquired by Ali, may be a better way to express this aspiration.
Finally, to reiterate, these two are very important. Equivalent to one is to evaluate the ability of software developers, one is for the effective control of collaborative process, but I just can do it when it comes to the big party a or party a, it has a good product manager, project manager, may be two people, also can be a person who, through this platform to complete the project. Hire a temporary software engineer for a short period of time, and then get the job done within the allotted time for the project, something like that.
The application prospect of artificial intelligence in insurance industry
Jian-yong fang
At present, the insurance industry in China mainly focuses on property insurance and life insurance. In the early stage, the People’s Insurance Company of China was divided into China Life Insurance Company and The People’s Insurance Company of China, which is the specific practice of the separation of property insurance and life insurance. However, after 2008, China Life Insurance Company set up an independent property insurance subsidiary, and People’s Insurance Company of China also set up a life insurance subsidiary, and we all returned to the one-stop service for customers.
Property insurance, hence the name incredible is property insurance, according to the kind of property can be divided into car insurance, housing insurance and other physical property insurance. Life insurance, is directly related with people, there are health, casualty, another is with the above two kinds of risks are set point side which has the function of savings insurance of share out bonus, such risks, the insurance company development fast do big propellant, such risks insurance function is weak, because carry correlation between larger and smaller gains with deposits, grew by another which has the function of investment’s share of the risks by extrusion.
How artificial intelligence can be developed in the field of insurance depends on the characteristics of insurance itself and the corresponding process. The primary intention of insurance is protection, which is based on the law of large Numbers and allows more people to spend less money to achieve a wide range of accident insurance coverage according to the actuarial result matching the premium and the insured amount. Therefore, whether it’s property and life insurance, insurance actuary design products this link is very important, it can directly determine whether insurance products in the exercise of safeguard function at the same time also have certain profit, the profit is the basis of maintain sustainable insurance company management, artificial intelligence is to help actuaries through data mining, combined with a proactive forecast of corresponding mathematical algorithm, designed a and a adapt to the current market demand for high quality insurance products.
Had insurance product, want to have enough customer to buy ability to go, otherwise also cannot form the foundation of large number that protects. Early to promote insurance company insurance products Numbers as the main strategy, China life insurance company, for example, in zhejiang province alone has more than thirty thousand insurance agents, their distribution in zhejiang province various counties urban district town, according to the corresponding practice administrative level sinking, establishing a branch, branch and office, organization, management fees and commissions spending is quite striking, these are the costs, will undoubtedly increase the cost of customers, also increase the company’s operating costs, thus affecting trade competitiveness.
The application of artificial intelligence in the promotion stage of insurance products is to enable insurance products to directly “face to face” with customers and eliminate all intermediary links. Now most insurance companies have opened the function of online insurance purchase, but the number of customers who take the initiative to buy from the corresponding platform is still very rare. On the one hand, the traditional business development habits of insurance companies themselves are not enough to reverse, do a little better, is the insurance agent to help customers in the online order; On the other hand, the online insurance channel is still not smooth, the cost of customer acquisition is still very high. More hot network insurance company on the market, is a three-horse (alibaba ma, tencent ma, ma ping an insurance) investment of the insurance, ali trading and payment basis, tencent has users and payment basis, ping an insurance is the insurance operation basis, all three investment insurance constructed Ann user + market + insurance products + pay insurance product purchase link, the whole flow of artificial intelligence here reflects the insurance product purchase closed loop without human intervention, it doesn’t require a lot of management personnel and the insurance agent.
The application of artificial intelligence in insurance fee collection is mainly manifested in two aspects. First, for the same amount of insurance, different customers charge different premiums, then how to judge this difference? If is one of the largest car insurance, you can pass the driving violations or be or get out of danger of automatic matching, if it is one of the life insurance, health insurance, the corresponding customer health conditions such as the history of medical records, medical records or ever be or get out of danger of automatic matching, this need big data to predict customer overall analysis and combining the actual situation of a single customer. A phase is produced each year pay insurance products through the credit card deducted automatically binding, as simple as make coal, water and electricity fee, don’t have to like usual make special insurance agent to maintain, to urge the customer pay cost, because doing that the drawbacks of the two, a situation in which all customer service levels, easy to cause the customer complaint, or premium for failing to pay guarantee slip invalidation; Another kind of circumstance is insurance agent misappropriate the premium that client pays, because traditional practice is insurance agent collects premium to client next unified hand in company to register.
The application of artificial intelligence in claim settlement is mainly manifested in image recognition and accurate claim settlement pattern matching. We set a scene, such as car insurance, the customer through the facial recognition on the network insurance company system, the responsibility to determine the written report, and pictures of the vehicle damage, the state of life or personal injury medical report sent to unattended online claims center, claims center by artificial intelligent automatic identification text and image, and then according to the customer’s insurance, matching claims automatically by the computer operation mode, instant give settling fee payment and the corresponding record deposited in the archives.
The application of ARTIFICIAL intelligence in customer service is mainly the function of customer assistant. The customer assistant is capable of speech recognition, including Mandarin, English and even other minor languages, as well as mainstream Chinese dialects. The customer assistant is also able to communicate with customers as an insurance specialist, which requires a large enough pool of experience to support it, ranging from general insurance talk to professional insurance experience. In addition, the most important part of artificial intelligence is the communicability of human-computer interaction. The customer assistant should find the accurate response mode of the corresponding customer.
Of course, these conclusions are based on my five-year working experience in Zhejiang Branch of China Life Insurance Company and my current understanding of ARTIFICIAL intelligence. This conclusion may have more room for improvement with the development of the insurance industry and the promotion of artificial intelligence.
A guide to advanced development of artificial intelligence
Jian-yong fang
Artificial intelligence was included in the report of the 19th National Congress of the COMMUNIST Party of China, followed by the establishment of Zhijiang Laboratory in Zhejiang Province, the Institute of Artificial Intelligence in Xidian University, and the Nansha Institute of Computer Vision in Guangzhou. The government and the academic community responded to this “national policy”. With the disappearance of China’s labor-intensive demographic dividend and the transition from a production-oriented society to a consumption-oriented society, artificial intelligence will play a very important role. On the one hand, it can make up for the shortage of labor population, and on the other hand, it also highlights the dividend of technology-based talents. And one more thing is that we need to point out, is the artificial intelligence field on a global scale, talent reserves accounted for 850000 in the United States, less than one over ten of the United States in China, the Chinese scientists involved in the global research and development of artificial intelligence is an important factor to drive forward China’s artificial intelligence, and cultivating talents of Chinese artificial intelligence is the most important mentor group.
It’s obviously not realistic to start from scratch, it takes at least eight or nine years for an undergraduate to be able to do independent research and development, but it also takes five or six years for a graduate student to be able to be part of a team in research and development. As one of the key points of the 13th Five-Year Plan, the participating groups of ARTIFICIAL intelligence obviously need transformational young and middle-aged talents in related fields, such as mathematics, computer and automation control. Speak like a zhejiang university professor, the first-class university teaching of artificial intelligence, second-rate university teach intelligent manufacturing, third-rate university teaching robot, we have a lot of people in society is the artificial intelligence, robot intelligent manufacturing is artificial intelligence, the ALPHA GO beat lee se-dol news event with artificial intelligence, which is one-sided.
Artificial intelligence involves many aspects. The most important ones are mathematical algorithms, machine learning, deep learning and computer vision. Of course, the most intuitive ones are probably some applications, such as speech recognition, natural language processing and image recognition. Iflytek has excellent performance in speech recognition. However, I recently learned that a youdao translation is also very good. It can not only realize speech recognition, but also realize speech recognition and language conversion of different languages. In terms of natural language processing, there are search engines Google and Bing. In fact, speech recognition is also an important application of natural language processing. Image recognition, is now more popular, as far as I know, a Stanford university Li Fei fly ImageNet, Beijing two startups is red, one is FACE++, a Thomson technology, but I feel like hikvision the $300 billion video surveillance of the listed company of the potential for more in the future, after all, it has and will have more image data as the training data.
Mathematical algorithms, machine learning, deep learning and computer vision are the foundation of artificial intelligence, and mathematics in particular is the foundation of the foundation. That is to say, there will be a huge demand for math talents from top universities in the future. Graduates from top math majors in China, such as Peking University, Zhejiang University, Fudan University, Tsinghua University, and University of Science and Technology of China, will make more than expected profits in the boom of ARTIFICIAL intelligence. Artificial intelligence is a big concept, machine learning is one of the pieces, deep learning is a piece of machine learning, computer vision is applied, and the mathematics is throughout the course of the basic framework of artificial intelligence, like many of math courses, such as higher algebra, mathematical analysis, differential geometry, functional analysis, complex analysis, probability and statistics, real variable function, complex function and numerical analysis courses, should be the cornerstone of the artificial intelligence system, 120 anniversary of zhejiang university, graduated from zhejiang university math department of shi yuzhu donated 5000 Wan Chengli big data research center, his meaning is very clear, Even with big data, you need core mathematical algorithms to support it, let alone those that don’t.
One of the hottest topics recently is deep learning, which is called deep neural network, and it is only in recent years that GOOGLE and FACEBOOK have made it to the top of the media. The founder of deep learning is HINTON, and the booster is LECUN. ANDREW NG is the most famous student in the Chinese community. His deep learning courses are very popular, and those who are interested can find them on Coursera.
In addition, talk about artificial intelligence production, study and research. In China there is a company that was established in 2015, called the heart of the machine, it is as an artificial intelligence all media, are the heart of lenovo’s entrepreneurial projects, after got the headlines today, morningside funds with investment, it has its own website, have their own WeChat public, also in the tiger sniff network, lei feng network, 36 kr, today’s headlines, zhihu since the media platform to open the heart of the machine column. My interest in it was mainly has three aspects, one is for the academic track, one is to interview the production of artificial intelligence, a is to organize some offline activities, translation of academic track is mainly on the arxiv about artificial intelligence, machine learning and computer vision in recent preprint papers, organizing offline activities I participated in the “CREATE WITH AI” baidu combat battalion. Hangzhou station, the heart of the machine as the role of the media support.
Once on WeChat, I talked with my undergraduate friend LOKI about the recent technological trends in silicon valley. Now he works in silicon valley and is quite familiar with the market. He said that big data would become a standard configuration and soon silicon valley would no longer talk about it. TensorFlow, an open source project on Github, is led by the GOOGLE BRAIN TEAM whose former director is ANDREW NG. TensorFlow, the 2.0 version of the DistBrief project hosted by ANDREW NG, is said to be the key to GOOGLE’s return to the Chinese mainland as the ai craze gets underway. From Internet COPY TO CHINA, TO mobile Internet, and then TO artificial intelligence, this is the third wave of the SPREAD of American information technology TO the Chinese mainland, and this third wave is just beginning.
It’s interesting to see that three years after Baidu hired ANDREW NG for $20 million a year, it launched its own third-party algorithm platform PaddlePaddle on Github, although the app hasn’t been popularized and the number of users is still small. In baidu’s organizational structure, the ARTIFICIAL intelligence department has been placed alongside the search department and the new project development department in the charge of Zhang Yaqin, and the ARTIFICIAL intelligence department has been put in the first place, indicating Baidu’s attitude. However, at present, the most widely used is GOOGLE’s TensorFlow, followed by FACEBOOK’s Github open source project Caffe2, TensorFlow is far ahead of Caffe2.
On the west side of Zijingang Campus of Zhejiang University in Hangzhou, hangzhou municipal government is building a concept called Chengxi Science and Technology Innovation Corridor. Now it has built the future science and technology City and China ARTIFICIAL intelligence town. It has first-class infrastructure construction, sufficient talent reserve and strong capital strength, just look at the high-end talents from the United States in place. We’ll see where the famous Alibaba headquarters will be located.
日攀科技 (机器海岸线) 