上海后花园论坛-上海419论坛-夜上海论坛

The Asian Cup is crazy again： Vietnam scored 11-0, the national football team broke the 9-year dilemma, and the sword refers to the World Youth Championship.

The Asian Cup is crazy again: Vietnam scored 11-0, the national football team broke the 9-year dilemma, and the sword refers to the World Youth Championship.

Text/Bin Shaokan Autumn (beginning) March, just as the Asian Football Championship started. The men’s soccer team participated in Uzbekistan U20 Asian Cup; It also brought many football festivals to Asian fans; A few days ago, the Asian Cup was crazy. First of all, Vietnam won two consecutive victories in the men’s soccer group match, beating Australia and Qatar. Unfortunately, they finally did it. Unfortunately, he was out again. Although he scored 6 points, he still couldn’t advance to the group stage.

In the first Asian Cup qualifiers of women’s football, the Australian women’s football team played crazy, beating Guam 13-0 and the China women’s football team beating the Filipino women’s football team 6-0; Another 11-0 victory hit the Singapore women’s football team hard. Although in the past, Vietnamese women’s football team was recognized as a weak team, with the development of Vietnamese football, their men’s and women’s football teams have risen a lot. In the future, they are bound to become the China team. Another new patient.

In terms of men’s soccer in China, in the group stage, 3 matches, 1 win, 1 draw, 1 loss and 4 points were accumulated. In the last round, thanks to the Japanese team’s victory over Saudi Arabia, the China men’s soccer team finally qualified in the group stage and reached the Asian Games. The top 8 in the youth competition, and that record also broke the dilemma of the youth national team for many years and returned to the top 8 seats after 9 years.

Judging from the overall process of the group stage, although the performance of China Men’s Football Team is remarkable, it does not have much advantage in the scene. Even against Kyrgyzstan, the weakest team in the group, the passport control organization of national youth players is completely at a disadvantage, and beating Saudi Arabia 2-0 is also a complete upset. In this game, the possession rate is as low as less than 30%. However, the game focuses on results. And then winning the bus is also commendable.

After the China men’s soccer team qualified, they also fell into an extreme situation. In other words, the South Korean team was the first to be criticized in the knockout. There are too many opportunities for men’s soccer in China. For China men’s soccer team, there is no way out. They can only fight hard to reach the top four, reach the top four of the Asian Youth Championship and win the World Cup. Tickets for the tournament, so the goal of China Men’s Football Team in this Asian Youth Championship is to beat the World Youth Championship with all one’s strength.

Bold! Ma Jing’s celebrity burst into controversy, VAR broke the double arrogant marketing, and C Ronaldo missed the grand prize from then on.

Since the VAR technology has been widely used in football, the voices of doubt have been one after another. Some people think that VAR makes the football game fragmented, which makes football lose a primitive aesthetic feeling, and even puts the manipulation of the game on the table. After all, when to watch or not to watch it is decided by the referee team, and things that are misjudged after watching it are sometimes staged in the Premier League. There is also a view that in most of the time, VAR still ensures the fairness and justice of the game, with less misjudgments and safeguards the interests of small and medium-sized teams. At least for Futre, a famous Atletico Madrid, VAR brings more positive aspects.

Indeed, considering the identity of Atletico Madrid, Futre is naturally a staunch supporter of VAR. At the peak of Atletico Madrid, they had at least one chance to win the Champions League, and this time the most likely opportunity was destroyed by Ramos’ offside goal. If there was VAR technology at that time, the first Champions League champion of Real Madrid in the three consecutive years would have ceased to exist. Of course, during the period of Real Madrid’s hegemony, because there was no VAR, there were many profits. In the game against Bayern, Luo Kaiyue, a notorious football stalk, was even born.

When he was a guest in the football column of O JOGO recently, Futre complained about the performance of the referee in Real Madrid’s two Champions League finals. He also talked about the recent referee incident in Barcelona. He said: "On the issue of referees, Real Madrid is the club with the least right to make irresponsible remarks about other teams. If Real Madrid asks La Liga officials to investigate the consulting fees paid by Barcelona to referees, then I suggest that all clubs in Europe should ask UEFA to judge the referees who have hosted Real Madrid’s Champions League matches.

In addition to this provocative speech, Futre also talked about some opinions on VAR. He believes that this technology is undoubtedly making football more fair, and it can provide some protection for the interests of small and medium-sized teams and reduce human errors caused by referees. Although the misjudgment still exists in the presence of VAR, the penalty made through VAR is correct most of the time, and these techniques significantly reduce the situation of benefiting from offside goals. Futre also said that if the VAR technology can be used as soon as possible, then the offside goals of Ramos and Cristiano Ronaldo in the Champions League should be cancelled.

Speaking of the performance of Messi and Cristiano Ronaldo in this World Cup, he is not optimistic that Cristiano Ronaldo will continue to play for the national team. He said: "Cristiano Ronaldo is a very good player, but the retrogression brought by age is inevitable. His mistakes are increasing every year, such as offside. Frequent offside is not the expression of goal desire, but the degradation of body and consciousness. With the capture of millimeter accuracy VAR, C Ronaldo’s offside times will only increase, but he is always at the forefront. He needs more offside to get a single shot. "

"Messi is not the same type of player. Messi has become more comprehensive. He is very smart. Messi is not a comparable object in terms of position or ability. However, Cristiano Ronaldo became a complete opportunist. The appearance of VAR and the decline of his body made his previous scoring ability unsustainable. After leaving La Liga, Cristiano Ronaldo won less honor. He was also replaced by players such as Mbappé in the competition for the highest honor in football, but Messi was still himself, and he was always compared by others. "

It is worth mentioning that Futre is not a rat. He is one of the legendary Portuguese superstars, the legendary winger of the national team, once the idol of Figo, and won the European Silver Ball Award in personal honor. He is a supporter of Messi. In this season’s Champions League match between Benfica and Paris Saint-Germain, Futre personally visited Messi through Campos’ relationship. He not only made a gesture of worship, but also hugged and took a photo with Messi, and finally asked for his autograph.

0-9 to 1-0, revenge! The penultimate madness of the Premier League： finally beating Liverpool and ending the seven-year wait.

At 20:30 on March 11th, Beijing time, in the 27th round of the Premier League, Salah missed a penalty and Billing scored the only goal in the game. In the end, Liverpool lost to Bournemouth 0-1 away and swallowed the first defeat in nearly 6 rounds.

Before this campaign, Liverpool scored 13 points in 5 rounds, 4 wins and 1 draw, and rushed to the fifth place in the league, only 3 points behind Tottenham Hotspur. They just washed Manchester United 7-0 at home in the last round. If they can score all three points from Bournemouth in this round, they will temporarily overtake Tottenham on goal difference and rise to the fourth place in the Premier League. Bournemouth lost 11 games in the last 15 rounds, only won twice, ranking the bottom of the Premier League with 21 points in 25 rounds.

In the first leg, Liverpool beat Bournemouth 9-0 at home, setting a record for winning the biggest score in the Premier League. Moreover, Liverpool have won seven consecutive victories against Bournemouth, during which they kept six clean sheets. The total score of these seven games reached a disparity of 28-1.

Who would have thought that all kinds of data predicted that Liverpool should take Bournemouth lightly, but in the end this round broke out with a 0-1 upset. For Liverpool, after losing these three points, it was doomed to not return to the top four after this round, but Burnley, who got the key three points, instantly overtook West Ham United, Leeds United, Everton and Southampton, not only got rid of the bottom of the league, but also escaped from the relegation zone and rose to 16th place. This is the first time that Liverpool lost to Bournemouth in the past seven years. The last time they lost to Bournemouth, they lost 3-4 away in December 2016.

Game review-

In the 29th minute, Ouattara made an inverted triangle cross from the bottom of 1V2 on the right, and scored a goal from Bilin Middle Road, with Bournemouth leading 1-0.

In the 69th minute, jota made a point, Salah took a penalty kick and flew wide, but Liverpool missed the equaliser and lost 0-1.

How to improve the navigation accuracy of autonomous robots？ Artificial intelligence： strengthen machine learning!

Robot navigation has become more and more advanced in recent years, but in very crowded environments, such as public areas or roads in urban environments, most robots still cannot navigate accurately. In order to be widely used in smart cities in the future, robots need to be able to navigate reliably and safely in these environments without colliding with humans or nearby objects.

Image source: Martinez-Baselga, Riazuelo & Montano

The collision trajectory of the robot trained by standard exploration strategy (left) and the successful trajectory of the robot trained by intrinsic reward in the same scene.

Researchers from university of zaragoza and Aragon Institute of Engineering recently proposed a new method based on machine learning, which can improve robot navigation in indoor and outdoor crowded environments. This method is used inarXivAccording to a paper published in advance on the server, it is necessary to use intrinsic rewards, which are essentially "rewards" that AI agents get when they perform behaviors that are not strictly related to the tasks they try to complete.

Autonomous robot navigation is an unresolved problem, especially in unstructured and dynamic environments, robots must avoid colliding with dynamic obstacles and achieve their goals.

Facts have proved that the deep reinforcement learning algorithm has high performance in success rate and time to reach the goal, but there are still many places to be improved.

The method introduced by Martinez Baselga and his colleagues uses intrinsic rewards, aiming at increasing the motivation of agents to explore new "states" (i.e. interactions with their environment) or reducing the level of uncertainty in a given scenario, so that agents can better predict the consequences of their actions.

In their research background, researchers specially use these rewards to encourage robots to visit unknown areas in their environment and explore their environment in different ways so that they can learn to navigate more effectively over time.

Compared with the same algorithm with ICM (intrinsic reward), the training index of the most advanced algorithm.

Most deep reinforcement learning for crowd navigation focuses on improving the processing of network and robot perception.

My method studies how to explore the environment during training to improve the learning process. In training, the robot does not try random movements or optimal movements, but tries to do what it thinks it can learn more from them.

The researchers evaluated the potential of using intrinsic rewards to solve robot navigation in crowded spaces in two different ways. The first one integrates the so-called Intrinsic Curiosity Module (ICM), while the second one is based on a series of algorithms called Efficient Exploratory Random Encoder (RE3).

The researchers evaluated these models in a series of simulations, which were run on the CrowdNav simulator. They found that their two methods of integrating intrinsic rewards are superior to the most advanced navigation methods of crowded space robots developed before.

In the future, this research can encourage other robotics experts to use intrinsic rewards when training robots to improve their ability to cope with unforeseen situations and move safely in a highly dynamic environment.

In addition, the two models tested based on intrinsic rewards will soon be integrated and tested in real robots to further verify their potential.

The results show that by applying these intelligent exploration strategies, the robot can learn faster and the final learning strategy is better. And they can be applied to existing algorithms to improve them.

In the next research, it is planned to improve the deep reinforcement learning in robot navigation to make it safer and more reliable, which is very important for using it in the real world.

The cost of obtaining customers is rising year by year. How can intelligent marketing grasp business opportunities and maximize clues？

The decline of mobile internet traffic dividend is a market consensus, and it has entered the stage of stock game.

As a result, many marketing practitioners have found a problem:

The cost of obtaining customers for enterprises is getting higher and higher, even to the stage of doubling.

The cost of customer acquisition has become the "sword of Damocles" hanging over the heads of marketers.

Harvard Business Review has studied the value retained by users and found that 70% of companies agree that it is cheaper to retain existing users than to regain a user; The CAC (customer acquisition cost) of a new user is more than five times higher than that of keeping an old user; Increasing the user retention rate can increase the profit by 25%~125%.

So, since the acquisition cost of new users is so high, can we give up new users?

Obviously impossible. At present, everyone’s approach is to grasp both new customers and customer retention.

A fiery new way has emerged-refined operation.

Regarding refined operations, some people think that users are diverted from the APP side of products to WeChat official account and WeChat communities; Some people think that it is to send greeting messages to users of offline stores every holiday; Others believe that it is the automatic reach of thousands of people. Whether or not, these practices are the embodiment of refined operation, but not the whole picture.

Refined operation refers to the behavior of combining the data analysis of channels, user behavior and other dimensions with the development stage of the enterprise to carry out targeted operation activities for users in order to achieve operational objectives. This operation goal can be to gain more new users, improve the retention rate of users, improve the conversion rate of core business links, etc., and it is necessary to pay attention to the whole life cycle of customers.

Traditional marketing of enterprises often wastes a lot of resources on invalid customers, and the labor cost and time cost of acquiring customers are getting higher and higher. Now the era of extensive operation and volume marketing has passed, and it has been replaced by refined operation. Studying the life cycle of users is the complete development process from the beginning of contact with products/services to the end of leaving products/services, and implementing different marketing strategies for different stages.

Refined operation requires operators to pay attention to the whole life cycle of customers. The word customer lifecycle management is not unfamiliar to marketers, but there are not many enterprises that really do a good job in customer lifecycle management. There are many reasons, such as unclear division of stage strategies and scenarios, lagging human resources, high operating costs, etc., which may affect the effect of user life cycle management. However, with the development of artificial intelligence and big data technology, marketing has also begun to move towards intelligence and automation. A set of automatic marketing scheme put forward by Wofeng Technology is a good help for enterprises to solve this problem.

Overview of outbound robot application scenarios

On the one hand, the quality of non-advertising leads is uneven, it is difficult to identify the intended customers, and the industry competition is fierce; But on the other hand, advertising in online channels is often accompanied by high advertising costs. How to quickly reach and identify the obtained clues is a good way to ensure that the clues are not lost.

Wofeng technology outbound robot can batch high concurrent outbound calls and automatically complete pre-sales clue filtering. For missed customers, it can automatically call back and reach users efficiently. The outbound robot can intelligently identify customer intentions, automatically judge and mark customer intentions, and improve the efficiency of clue screening. AI algorithm intelligently extracts key information expressed by customers and generates customer portrait labels. Enterprises can conduct differentiated marketing according to the labels and provide decision optimization for operators.

Traditional manual communication and drainage methods are inefficient, ineffective and costly, and it is difficult to reach all customers. The customer label is not clear, and it is difficult to do hierarchical and refined operations.

Enterprises can use the customer database to launch intelligent group calls to public customers, screen high-intention customers to invite them to add, and quickly improve the pass rate of adding powder. Enterprise WeChat powder application is different from a micro-application, which will not appear on the regular "new friends" page, but will be displayed in the notification message of enterprise WeChat. Simply using active powder addition, users can easily ignore it, resulting in low powder addition rate. Combined with AI outbound, users can be informed of the specific entrance to apply through friends, thus improving the powder adding rate.

In the event marketing scenario, the online delivery cost is high, and it is impossible to accurately reach consumers point to point; SMS, Push and other methods are passive and have low opening rate.

1) Accurate customer identification and reach: distinguish registered customers, historical transactions and other customer types, generate customer portraits based on historical consumption information, customer registration information and outbound history, and identify core target customers to accurately push marketing activities.

2) Speech customization: preset multiple sets of standard speech according to common event marketing scenarios, and form a differentiated speech library in opening remarks, brand wake-up words and temptation points (gifts, discounts, coupons, etc.). Match the customer’s portrait accurately, and ensure the transformation effect.

When a customer initiates an order, but has not paid for a long time, or has paid the deposit, but has not paid the final payment, etc., the enterprise can use AI outbound to make a reminder, so that interested customers can pay quickly.

During the promotion period, under the gameplay of deposit inflation, users need to pay deposit+final payment, but the date of final payment is different; Enterprises can call for payment on each user’s last payment day, and attach a page link through SMS, which can greatly improve the efficiency of payment collection.

It is difficult to quantify the work of service personnel, and the efficiency of manual sampling and return visit and manual entry of results is low.

After the sale, the outbound robot can make outbound calls in batches with high concurrency, quickly complete the return visit task, conduct targeted marketing for customer problems in the return visit process, and automatically recall customers who are not connected, ensuring the reach rate and return visit effect. After the return visit, statistical analysis report and public opinion analysis results can be automatically generated according to the call records, which can assist business decision-making and provide reference for the next marketing of enterprises.

Wofeng technology outbound robot runs through the whole life cycle of users from five scenarios: clue screening, private domain drainage, activity notification, order reminder and after-sales service. For enterprises, outbound robots, as intelligent tools, help enterprises to operate, improve user retention rate, reduce costs, improve transformation, and continuously tap the growth value.

Plus One Robotics received $50 million in Series C financing.

Plus One Robotics is an AI vision software and robot package handling solution provider, which provides artificial intelligence software, equipped with terminal robot gripper, and provides the perception and operation required for picking and placing packages. Recently, Plus One Robotics announced the completion of the $50 million Series C financing. The global venture capital company in this round of financing was led by Scale Venture Partners, and Tier Capital Partners, Tyche Partners, ROBO Global Ventures, Translink, McRock and Pritzker Group Venture Capital participated in the investment.

Promoting learning by competition and cultivating artificial intelligence learning thinking for teenagers

How high is the learning threshold of artificial intelligence? Is it necessary to have a programming foundation? Can teenagers learn artificial intelligence? For these questions, some schools have given answers. At present, many schools such as high school affiliated to renmin university of china, Beijing Dewei British International School and Beijing Haidian Foreign Language Experimental School have launched artificial intelligence learning courses.

In an interview with reporters, Yuan Zhongguo, head of the Information Technology Teaching and Research Group of high school affiliated to renmin university of china, said: "We deeply understand the important role of artificial intelligence education in improving students’ innovative thinking and practical application ability. At the same time, in the long-term teaching practice, we also found that the teaching method combining theory with practice is more attractive to students and promotes the improvement of students’ practical ability. "

In fact, as early as 2020, the Ministry of Education issued the Key Points of Education Informatization and Network Security in 2020, pointing out that the construction, application and promotion of artificial intelligence education courses in primary and secondary schools should be continued. Release the curriculum package of artificial intelligence education in primary and secondary schools (junior high school edition and senior high school edition) and support service system and promote its application. Promote the development, trial application and popularization of the curriculum package for primary schools.

For schools, lowering the learning threshold, promoting students’ interest in learning in a way that teenagers can understand, and setting reasonable courses and teaching materials are the key points in the teaching process. In this process, the school is also seeking cooperation with enterprises, hoping to introduce more suitable educational methods and first-line industry experience to enhance students’ interest in artificial intelligence. Amazon DeepRacer is also an attempt to enter the teaching field of vision under this background.

It is understood that Amazon DeepRacer is a 1/18-scale fully-automatic racing car driven by reinforcement learning and 3D racing simulator launched by Amazon Cloud Technology in 2018. It has the characteristics of easy entry, understandable principle and simple operation, and can give attention to both fun and let operators systematically understand the knowledge of reinforcement learning.

Amazon DeepRacer is characterized by low threshold and easy operation. For teenagers, you don’t need the foundation of programming learning to get started. Gu Fan introduced: "We only need to train the model with some parameters, and then deploy it to the car. The camera of the car will judge the actual situation of the track and use its trained method to judge where to drive and at what speed."

The deployment principle of Amazon DeepRacer is the core algorithm reinforcement learning of machine learning, which is consistent with the algorithm principle in practical work. At the same time, in the process of operation, it is necessary to prepare data, train the model, and then deploy the model. If the deployment effect is not good, the model parameters need to be optimized and iterated again. This process is consistent with the process of helping customers in various industries solve problems related to machine learning in their daily work. In addition, the overall experience is consistent. If it is not deployed properly, it still faces the situation that the car cannot follow the predetermined trajectory, that is, the solution fails.

Therefore, Amazon DeepRacer actually trains young people’s artificial intelligence thinking through racing. Yuan Zhongguo said: "The introduction of Amazon DeepRacer provides students with an opportunity to experience from algorithmic intelligence to entity intelligence. For primary and secondary school students, learning algorithmic intelligence is very attractive. We are very pleased to see that children are still enjoying training Amazon DeepRacer after class. Therefore, the introduction of good platforms and products can play a very positive role in promoting students’ learning artificial intelligence.

Since 2020, Amazon Cloud Technology has successfully held a series of education and training activities in China, attracting more than 50,000 people to actively participate. In 2023, Amazon Cloud Technology plans to hold more than 100 online and offline events, covering hundreds of thousands of people. These include Amazon DeepRacer China League for developers and machine learning enthusiasts, various DeepRacer industry leagues, enterprise competitions and trainings for different industries and enterprises, and a series of activities for teenagers jointly organized with partners.

At the kick-off meeting of the 2023 Amazon DeepRacer China series, Amazon Cloud Technology announced that it had reached a strategic cooperation with the Science and Technology Education Collaboration of China Education Association and the Shanghai Artificial Intelligence Industry Association. It will focus on Amazon DeepRacer’s artificial intelligence learning tools and invest resources in four aspects: curriculum development, teacher training, platform support and competition-based learning, so as to provide relevant training for 100,000 China teenagers in the next three years and promote the development of artificial intelligence education for teenagers.

Elaine Chang, vice president of Amazon Global and executive director of Amazon Cloud Technology Greater China, said: "Amazon DeepRacer has low barriers, strong interest and high operability. We hope to use this unique learning tool to develop systematic courses and launch rich education and training activities, lower the threshold for people to learn artificial intelligence, stimulate their interest and enthusiasm for machine learning, and help cultivate artificial intelligence talents at all stages. "

This combination of enterprise experience and education and training is also expected to promote the early education process of young people in the field of artificial intelligence. Yang Nianlu, executive vice president of China Education Association, said: "Teenagers are an important part of talent training, and the educational concept and mode of talent training should also keep pace with the times. Artificial intelligence education can cultivate students’ logical thinking, innovative ability and practical ability, which is of great help to improve students’ comprehensive quality and stimulate teenagers’ curiosity, imagination and desire to explore. It is hoped that through the joint efforts of the three parties, the artificial intelligence education for young people can be brought to students in a more universal way, and the scientific literacy of young people can be gradually improved to consolidate the foundation of scientific and technological talents in China. "

According to "White Paper 2020 on the Integration of Artificial Intelligence and Manufacturing Industry" issued by the National Research Center for the Development of Industrial Information Security in 2020, there is an artificial intelligence talent gap of 300,000 in China. With the rapid development of the industry in recent years, this gap has grown geometrically. Zhong Junhao, secretary-general of Shanghai Artificial Intelligence Industry Association, said: "Talent is an important foundation for the development of artificial intelligence technology. Introducing artificial intelligence education in adolescence is very important for cultivating innovative talents. "

Bayern has scored 51 consecutive home games …

Bayern has scored goals in 51 consecutive home games, ranking third in the Bundesliga in terms of consecutive home goals.

In addition, Bayern had scored in 87 consecutive Bundesliga games before losing to augsburg in the seventh round of this season. On September 17th, 2022, Beijing time, Bayern lost 0-1 to augsburg, and this data was also terminated.

# Bayern # # Bundesliga #

A ＂mobile＂ postdoctoral fellow came to the research station of the Big Data Center.

▲ The research results of Shanghai Big Data Center continue to empower "one network to run". Our reporter Xing Qianli photo

Wang Qian has two identities: "Dr. Wang" and "Wang Gong".

This stems from a special "cross-border". She entered Post-Doctoral Research Center, a big data center in Shanghai through various selections. During her two-year postdoctoral career, she participated in many major digital technology projects in the city. Recently, she is connecting with emerging technologies such as "AI+ blockchain", and the technical research on blockchain smart contracts will come to a successful conclusion.

Talent is the foundation and support of innovation. Shanghai is accelerating the construction of high-level talent highland and exploring the effective linkage between the strategy of rejuvenating the country through science and education, the strategy of strengthening the country through talents and the strategy of innovation-driven development.

At the end of last year, Shanghai Guangfa issued the "Hero Post" and "Recruitment Order" and took out 5,157 post-doctoral positions. City Big Data Center is one of them, attracting talents with full chain service.

-Introduce talents-

Combine the needs to attract "fine" talents

As early as November 2020, Ministry of Human Resources and Social Security and the National Postdoctoral Management Committee approved the establishment of the Shanghai Big Data Center Post-Doctoral Research Center, becoming the first provincial and municipal big data management institution in Post-Doctoral Research Center. In 2021, it was officially listed and issued a "recruitment order" to the society. Two postdoctoral researchers, including Wang Xi, were recruited in the first phase. "Cultivate practical professional big data talents based on business needs. Fully combine the actual work of the center, create’ fine’ talents, let them realize the cross-promotion of theory and practice, grasp and understand macro policies, and make a qualitative leap in the ability to evaluate new technologies, "said the relevant person in charge of the Municipal Big Data Center.

Attracting talents with "industry" is the secret of the city’s big data center. On the one hand, the recruited postdoctoral fellows can deeply participate in the top-level design of Shanghai public data and a network office, focusing on the research closely related to urban operation; On the other hand, it has the opportunity to participate in key R&D plans and major scientific research projects at or above the provincial and ministerial levels, and participate in the formulation of relevant national and local standards.

Digital economy is a new industry facing the future, which has great potential and application development space. Recently, the city’s big data center combined with the focus of digital economy development, added two research directions: trusted circulation of data elements, time series analysis and machine learning, and recruited postdoctoral staff again. With the deepening of research results, we will continue to recruit talents on demand.

-Yucai-

Work goes hand in hand with further study.

Connecting innovation chain with the talent chain, and promoting the energy level of industrial chain.

When Wang Qiangang "entered the station", she was confused: she didn’t have much research on blockchain technology, entered the postdoctoral workstation, and faced a number of key projects.

At that time, Zhu Junwei, the co-director of the workstation and deputy director of the Municipal Big Data Center, outlined the future plan, and Jin Yaohui, the co-director of the mobile station and a professor of Shanghai Jiaotong University, organized a special seminar. After more than one month of selection and inspection, she decided to participate in the key project-the construction of the blockchain electronic material library of "One Netcom Office" in Shanghai. "It not only fully respects my wishes, but also ensures that my research has application scenarios, avoiding the disconnection between theory and application, which is of great benefit to later research."

Work is "further study". After mastering the basic situation, she joined the team directly and became a core member. This year, she focused on how to make smart contracts more "smart". She has "broken" the whole operating mechanism of the blockchain smart contract, such as triggering, transmitting, verifying, updating and network-wide consensus, and tried to construct multi-dimensional perception and interaction of users, chain environment, contract set and even offline applications, so that the contract can become a "nervous system" that can touch the whole blockchain ecology, thus upgrading the contract program from "running" to "running".

In order to improve the ability of chain management and control, Wang Xi has also launched an independent contract-driven self-monitoring scheme of "managing chains by chains". Combined with artificial intelligence, she overcame the three-stage modular call scheme of "before, during and after the chain" of "AI+ intelligent contract (AI-SC)" in her research, and successfully upgraded the intelligent contract to a smart contract.

Learn new technologies in practice, and find new scenes and new paths from new technologies. The good experimental ecological field in the station gives Wang Qian a huge space to display her talents. She also completed the construction of the "One Netcom Office" electronic materials alliance chain system, and participated in the establishment of the Shanghai local standard of "Technical Specification for Sharing and Application of Electronic Materials". This means that the citizens’ materials have been submitted once, reused many times and shared by the whole network. "Constantly explore cutting-edge technologies to solve practical application problems and truly benefit every citizen."

-Use talent-

All-round "escort"

The subjective initiative of all kinds of talents can not be separated from the "flowing water" in which one party loves, respects and uses talents.

Post-Doctoral Research Center, the city’s big data center, has built an integrated scientific research organizational structure system from technical foundation, institutional support to application services, forming a complete industrial chain of scientific research. At the same time, we have long-term cooperation with many leading enterprises and well-known experts and scholars in the industry, and inserted "wisdom wings" for talent training. When postdoctoral researchers encounter difficulties in their research, the workstation organizes a "Central Postdoctoral Salon" to invite experts inside and outside the center to answer questions and collide with new scientific research ideas and methods in mutual discussion.

The perfect talent assessment and evaluation system has fully released the research potential of researchers. During the postdoctoral research, the workstation carries out examination and evaluation at the beginning, middle and exit stages; Explore the mode of "regular briefing" to dynamically reflect the progress of postdoctoral research in the station. Involving postdoctoral research work, whether it is achievement declaration, research funding application or title evaluation, you can get relevant support. Wang Xi said that it was such an all-round "escort" that she completed two applications for invention patents and published two papers in the first year as a postdoctoral fellow, and she gained a lot. "I also hope to promote digital technology to radiate the Yangtze River Delta region and even the whole country."

The site also solves worries for researchers. Not only help to apply for talent apartments, but also pay bonuses according to performance appraisal, apply for Shanghai household registration, etc., "so that more postdocs can concentrate on their research fields with peace of mind."

Artificial intelligence is ＂windy＂, how to make steady progress in the data labeling industry？

Artificial intelligence is "windy", how to make steady progress in the data labeling industry?

Recently, AI (artificial intelligence) has once again ushered in a wave with the fiery chat bots, and it is even called "it will trigger the next industrial revolution". During this period, the field of data annotation has also appeared in various fields, such as the high valuation of data annotation platforms, the surge in demand for text companies, and the beginning of the split of data annotation teams by AI manufacturers … To describe the data annotation industry in a network term, it is once again.

More demanding data labeling requirements

"Artificial intelligence will change the world, so who will change artificial intelligence?" This is the question of AI scientist Li Feifei. Fifteen years ago, the AI ? ? community generally believed that better algorithms could bring better decisions, but it was discovered in the real application process:

The breakthrough of the algorithm’s trajectory to the deep neural network has led to many application scenarios, and then more data can be used for training, and the computing power will be cheaper because of its mass production. The above process will also promote the algorithm’s further improvement. On the contrary, some industries have not produced data that can be used for training, which leads to the stagnation of algorithms and computing power in this field.

Algorithm, data and computing power complement each other and restrict each other.Fortunately, Li Feifei realized 15 years ago that even the best algorithm can’t be used without good training data that can reflect the real world. Therefore, in today’s explosion of Chat GPT, people not only pay attention to its algorithm, but also realize.In the wave of large model trainingThe demand for traditional data annotation is likely to decline, but at the same time, it isBring another higher demand for data annotation.

The "Obstacle" of Data Labeling Development

Data annotation is a very important part in the field of artificial intelligence, but it has not been established yet.Sound industry standardsTherefore, it is marked that the competition between companies is blindly fighting for low prices, and Party B who finally gets the project is often unable to undertake it. The chaos in the industry makes it difficult for labeling companies that survive in the cracks to develop in a positive direction, and the internal friction of the whole industry is tantamount to self-destruction.

At the same time, there are serious problems such as insufficient professional knowledge background, low understanding of the industry and low labeling efficiency, which are also obstacles to the development of data labeling industry. However, if we ignore the professional ability of data annotators, there will be a growing gap between the data annotator industry and the overall development trend of artificial intelligence. So, in the face of the difficulty of the development of data annotator industry, where should we start?

Starting with the cultivation of talents, it is a key step for enterprises that want to start a business in the field of AI data labeling to do a good job in the incubation of artificial intelligence trainers.

Industry development is difficult, and talent training is the foundation.

As we all know, the overall development of artificial intelligence industry is rapid, and the application fields and scenes are becoming more and more complex. Therefore, simple labeling work such as making a box and marking a point will soon be replaced by AI, and the labeling work in the future will only become more and more professional and complex. ChatGPT is a typical example, and people have realized AI education AI.

But in any field, talent training is always the first. For AI training, the quality of data labeling is of great significance. If there are inaccuracies or even errors in the labeling process, it is likely to lead to very serious consequences. Even if fully automatic AI tagging is realized in the later stage, the data annotators still need to participate in the auditing process, so the importance of AI artificial intelligence trainers as data annotators is self-evident.

When we want to enter the field of data annotation, we need to pay more attention to the relevant training of data annotation talents.

Personnel training of artificial intelligence trainers

Speaking of this, some people will think: I understand that talent training is actually a new model in the development process of the data labeling industry. If you want to start a business and prepare for team transformation, you need to find expansion projects for the company, and it is completely no problem to consider the project about the incubation of data labeling artificial intelligence trainers. Then, what if individuals want to become artificial intelligence trainers?

No problem, the purpose of the artificial intelligence trainer talent incubation project is to cultivate excellent data annotation talents, build excellent annotation teams, provide a systematic talent training system for the industry, and provide a channel for data annotators to continuously improve their professional skills. Therefore, this project has a cooperation entry point suitable for both individuals and teams.

Finally, I hope that more enterprises and individuals who are interested in the labeling business can pay attention to and join the artificial intelligence data labeling industry, and jointly participate in the development of talent training and recommendation industry in the data labeling industry.