There is currently no content to display
Please add data record on website background.
AI Leads, Empowers Infinite Scenarios | The 4th Global Entrepreneur Conference was successfully held
分类导航 Nav

AI Leads, Empowers Infinite Scenarios | The 4th Global Entrepreneur Conference was successfully held

  • Categories:会议新闻
  • Author:
  • Origin:
  • Time of issue:2020-12-24 08:43
  • Views:

(Summary description)On August 28, the Global Entrepreneur Conference (GAISC), a concurrent conference of Shenzhen (International) Artificial Intelligence Exhibition organized by Shenzhen Artificial Intelligence Industry Association, was grandly opened in Hall 7 of Shenzhen Convention and Exhibition Center.

AI Leads, Empowers Infinite Scenarios | The 4th Global Entrepreneur Conference was successfully held

(Summary description)On August 28, the Global Entrepreneur Conference (GAISC), a concurrent conference of Shenzhen (International) Artificial Intelligence Exhibition organized by Shenzhen Artificial Intelligence Industry Association, was grandly opened in Hall 7 of Shenzhen Convention and Exhibition Center.

  • Categories:会议新闻
  • Author:
  • Origin:
  • Time of issue:2020-12-24 08:43
  • Views:

On August 28, the Global Entrepreneur Conference (GAISC), a concurrent conference of Shenzhen (International) Artificial Intelligence Exhibition organized by Shenzhen Artificial Intelligence Industry Association, was grandly opened in Hall 7 of Shenzhen Convention and Exhibition Center. This year's Entrepreneur Conference was held under the theme of "AI Leading, Empowering Infinite Scenarios". The guests of this year's GAISC included Xiao Jing, Chief Scientist of Ping An Group and Rotating Chairman of Shenzhen Artificial Intelligence Industry Association; Shen Linlin, Pengcheng Scholar and Director of Shenzhen University's Computer Vision Research Institute; Xu Tailin, Professor of Shenzhen University, Guangdong Artificial Intelligence and Digital Economy Laboratory (Shenzhen); Zheng Wenxian, Vice President of Shenzhen Yuntian Lifei Technology Company Limited; Yan Jingyu, Executive Vice President of Star River Industry Group and Partner of Star River Capital Ltd., Guo Zhen, General Manager of Guangdong Qinzhi Technology Research Institute Co., Ltd., Hu Lei, Founder and CEO of Beijing Dilu Shenwei Technology Co.

This year, the new infrastructure and the artificial intelligence included in it are the high-frequency words mentioned by the representatives of the two sessions. In the context of the accelerated arrival of the era of artificial intelligence, the new infrastructure construction with artificial intelligence as the core will usher in great development and is expected to become a strong engine for the development of the digital economy. With the introduction of new national infrastructure policies, and the deep integration of artificial intelligence and economic development over the years, the smart economy has become a major trend. The theme of this year's Global Entrepreneur Conference is "AI Leads, Empowers Infinite Scenarios", with two professional forums and one roundtable discussion.



At 10:00 a.m., Xiao Jing, chief scientist of Ping An Group and rotating president of Shenzhen Artificial Intelligence Industry Association, gave the opening speech of the conference. Xiao Jing mentioned: the last two years, due to the limitations of application scenarios, artificial intelligence technology from the high-speed growth period to the rational development stage, so that companies consider more artificial intelligence products to land and commercialization, more focus on creating value for users. And during the epidemic, AI's intelligent voice, face recognition, image recognition, self-driving and other technologies play an important role in virus research, epidemic prevention and control, information dissemination and other aspects, so that we can actually feel the benefits brought by AI technology. However, we should be soberly aware that China is not yet a strong country in artificial intelligence technology. Many technologies have not yet mastered in their own hands, for this reason, we need to increase innovation, accelerate the construction of artificial intelligence industry platform, to promote technology, application and service innovation in all aspects.

Shenzhen Artificial Intelligence Industry Association, as a bridge and link between the government and enterprises, adheres to the concept of sharing worries for the government, solving difficulties for enterprises and serving the industry, actively guiding AI enterprises to seize the opportunities of the digital economy era, rapid development, strengthening exchanges and cooperation with the government and enterprises, and promoting AI enterprises to play a greater role in the development of the real economy.

With the theme of "AI leads and empowers infinite scenarios", this year's GAISC explored how AI can play its own advantages under the current situation, help smooth the domestic circulation and focus on economic restructuring, and come out of a sustainable development path of AI.

After Xiao Jing's speech, the morning session of the Entrepreneur Conference officially started. The morning session focused on four topics: "biosensor, new infrastructure + AI industry, AI investment concept, 3D CV camera full-stack technology".

Xu Tai Lin: "AI-Biosensor opportunities and challenges



(Professor Xu Tai Lin, Guangdong Artificial Intelligence and Digital Economy Laboratory (Shenzhen) and Shenzhen University)


Professor Xu Tai Lin from Guangdong Artificial Intelligence and Digital Economy Laboratory (Shenzhen) and Shenzhen University shared the theme of "AI-Biosensor opportunities and challenges". Xu Talin talked about: the development of artificial intelligence has gone through several waves, from Alpha Dog's victory over Li Shishi, to Google's intelligent driverless car, 5G, big data and the development of biosensing, whose development trend is miniaturization, intelligence, integration and diversity. Why are smart biosensors an opportunity for development in China? First of all, the United States released a 20-year artificial intelligence development program, which points out that the intelligent medical concern for human health as the first, China's State Council issued a program to promote the development of Internet + medical opinions, clearly pointing to the Internet plus medical as an important development direction, the main development direction mainly supports 3D, wearable intelligence, intelligent manufacturing, artificial intelligence technology and intelligent medical robotics and other aspects.

We propose to the whole intelligent biosensing network, mainly including three parts. First, signal collection; second, signal transformation; third, AI data processing. We also foresee some functions of intelligent sensing, the first one is self-diagnosis, the second function is big data processing, and the third function is self-learning/adaptive, just like in remote diagnosis, medical wear, and home diagnosis play some such roles.

One of the main contents of our biosensing research is to develop set solutions for low content biomarkers in clinical aspects, the second is to combine new technologies, such as the Internet of Things, big data and other artificial intelligence technologies, to achieve artificial intelligence biosensing self-learning, self-assembly and self-adaptation. The third is the use of biosensors for cancer, and major diseases, such as Alzheimer's, as well as the early diagnosis of chronic diseases and health detection, such as the development of flexible electronic mimic devices, implantable sensing materials and devices. The fourth is the fusion of multidimensional perception, the development of key technologies for human-computer interaction. We also predicted several future sensor applications, one is intelligent wearable commercial sensors, the second is intelligent edible, digestible sensors, such as smart capsules, the third is based on the eye sensor, in addition, in the intelligent AI recognition of biological sensors mainly include fingerprint biometrics, face biometrics, iris biometrics and user identity recognition. 

Zheng Wenxian: "New Infrastructure Era, AI Industry Accelerates Landing

(Zheng Wenxian, Vice President of Shenzhen Yuntian Lifei Technology Co.


Mr. Zheng Wenxian, Vice President of Shenzhen Cloudflare Technology Co., Ltd. gave a speech titled "The Golden Age of AI Industry Landing under the New Infrastructure Wave". In his speech, Mr. Zheng mentioned that AI industry can be divided into basic layer, technology layer and application layer. At present, the development of AI in the application layer is the focus of industry attention, and the scale of the industry in the application layer far exceeds that of the basic layer and the technical layer. According to Zheng, three major elements are indispensable for AI to achieve industrial landing: first is the core technology, second is the focus on the scene, and third is the landing experience.

In terms of core technology, Yuntian Lifei insists on building a full-stack chain of chips, algorithms and big data; AI chips carry algorithms, and various business scenarios need the empowerment of AI chips.

In terms of scenario implementation, Yuntian LiFei has combined its strengths to focus on three major areas: public security, urban governance and new business, and has built the first dynamic portrait system, developed various algorithm platforms around Shenzhen criminal investigation, audit and network investigation, anti-trafficking and anti-drug, and developed a series of products based on artificial intelligence technology for shopping centers and chain supermarkets.

In terms of large-scale applications, CloudTek has been gradually moving from Guangdong, Hong Kong and Macao Bay Area to the whole country since 2015, developing a series of products and solutions around safe city, new business, AIoT, etc. So far, the products and solutions have landed in more than 100 cities nationwide, and the algorithm chip-based products have also reached strategic cooperation with Jingdong, Ali and Haikang.

About the "new infrastructure" to artificial intelligence industry opportunities, Zheng Wenxian think mainly in the following aspects: first, to solve the data transmission problem, 5G large bandwidth, low latency, high-speed technology features to solve the data transmission problem; second, arithmetic support, chip, data center to solve the arithmetic problem; third, the industrial Internet as a representative of a new generation of infrastructure technology, will bring new scenarios for the artificial intelligence industry to land, new exploration.


Yan Jingyu: "Great wisdom, great courage - artificial intelligence investment concept sharing


(Yan Jinyu, Executive Vice President of Star River Industry Group and Partner of Star River Capital)


Mr. Yan Jianyu, Executive Vice President of Star River Industry Group and Partner of Star River Capital, shared Star River Capital's investment philosophy. Yan said, Star River has been involved in many industries and equity investments since 2010, and now also established Star River Capital, so far Star River Capital has invested in a total of 130 projects, there are 30 listed companies, the success rate of listing is about 16-17%.

As an investor, Star River has invested in many artificial intelligence industry companies in the past decade, so we have some experience in understanding industry research.

From the investment point of view, Star River as an investor, more important to several issues. First, the accuracy of the algorithm. Artificial intelligence applications for algorithm stability and accuracy may be a more difficult breakthrough problem, especially in some of the human life safety or property security applications, which is a relatively large limit. Can errors be judged? Can there be an explanation mechanism in advance to predict whether it will be wrong, that is, its interpretability. Third, look at the ontology, only in the ontology has a very good foundation, and then through artificial intelligence to empower will do better. For example, there may be a lot of companies in the sweeping robot this academic background will be stronger than Coors, but the Coors sweeping robot can be in this field in the first echelon in the country, mainly because it is originally from the vacuum cleaner.

Household Lei: "3D CV camera, across the scale of application, 3D redefines the intelligent world


北京的卢深视科技有限公司创始人兼 CEO户磊

(Founder and CEO of Beijing Tilu Deep Vision Technology Co.


Ltd. founder and CEO of Beijing's Lu ShenShi Technology Co., Ltd. said that 3D is actually a more accurate application technology than 2D visual recognition. From the technical composition of the above, 3D and 2D have a little difference is that, in addition to the usual data, algorithms, it also has a 3D camera sensor evolution, data, algorithms, sensors eventually through a variety of different forms of modules or solutions, the formation of large-scale applications. With the maturity of 3D technology, the batch landing production makes the cost lower, with better cost performance, 3D visual perception system will become the standard eye of the machine.
High-precision 3D face recognition requires a lot of data training in the early stage, and unlike 2D images, 3D images must rely on 3D camera sensors to generate high-precision image data for basic data accumulation. Specifically, the Lu deep vision to provide three-dimensional full-stack technology, in the image machine module design, independent control of depth camera system design, platform selection, algorithm adaptation and other aspects, and in the temperature compensation, structure compensation, high-precision calibration and calibration and other aspects of the core technology, through the above technology, can reduce the requirements of device specifications and consistency, reduce assembly accuracy and cost, and improve the yield, and ultimately achieve low cost.

Now, 3D sensing technology and camera gradually enter the "scale application" stage, across the "scale application" gap, on the one hand, manufacturers need to continuously polish the product, innovation user experience. On the other hand, the 3D vision industry chain is long and immature, there is an urgent need to form a unified standard and consensus in the industry, these standards need to cover: 3D image evaluation standards, 3D camera parameter standards, 3D data interface standards, etc. The establishment of three-dimensional data standards and evaluation scoring system is the basis for three-dimensional applications, the professionalism and effectiveness of the standards and evaluation scoring system determines the effect and indicators of three-dimensional applications. The Lu deep vision for various data categories, are provided with data quality requirements level evaluation standards. According to the test of each depth camera commonly available in the market, the formula for calculating the overall depth map quality score of depth cameras is given by TLLI.

Depth quality score = F(accuracy, goodness of fit, continuity, null rate...)
At present, all the home-made 3D CV cameras developed by TILO have been mass-produced, with an error of less than 1mm in the 5m range, surpassing the international 3D camera giants and leading the world in accuracy, with exclusive patented technology, highly independent and controllable, and extremely high cost performance. The high-precision RGBD camera (standard type) can be used in face payment terminals, bank ATMs, unmanned cargo counters, subway face gates, AR/VR, item volume measurement and other scenarios; the 3D-Face ID smart module can be used in home smart door locks, smart safes, hotel apartment smart locks, smart access control and other scenarios.

The afternoon session mainly focused on the topics of AI arithmetic, GAN face changing technique, AI data, smart city and other fields.

Guo Zhen: "Make the best use of arithmetic power, embrace intelligence - AI era needs AI's special chip



(Guo Zhen, General Manager of Guangdong Qinzhi Technology Research Institute Co.


Guo Zhen, General Manager of Guangdong Qinzhi Technology Research Institute Co., Ltd. brought the theme of "Make good use of computing power and embrace intelligence". Guo Zhen said: the layering of artificial intelligence can be divided into four levels, chip layer, system layer, algorithm layer, application layer. Now many domestic AI technology development is more in the algorithm layer and application layer, AI technology is actually more lack of development in the chip layer and system layer, so Qin Zhi is currently doing things more inclined to the chip layer and system layer. Although CPU can do AI and deep learning, it still has some defects. If the traditional CPU is used to execute AI algorithms, but the speed is slow and the performance is low, so it is not commercially available. If the GPU used arithmetic power enough, but its energy consumption is too large, like the car battery currently can not support such a powerful energy consumption. So this example illustrates from a practical point of view why a dedicated chip for AI is needed. In addition, the CPU and GPU still belong to the category of general-purpose computing, no real AI-specific computing to do specific optimization. Therefore, NPUs emerged later, and the NPU principle is to simulate human neurons prominently in the circuit layer.

In addition, from the supercomputing aspect, we can see that: from the technical architecture, traditional supercomputing mainly uses CPUs, and AI supercomputing mainly uses smart chips, and from the application field, traditional supercomputing mainly faces scientific computing and engineering computing, and AI supercomputing mainly faces intelligent + industrial applications, and takes into account scientific research. Therefore, traditional supercomputing and AI supercomputing are not substitutes for each other, but rather complementary to each other.

Many of the limitations of artificial intelligence are mainly by three aspects. Limit one: limited generalization ability. I personally understand generalization ability is cross-domain cognitive ability, cross-domain perception or cognitive ability, for example, investment needs to be friends with time, put a lot of human reasoning on top of investment is also applicable, this is generalization ability, lack of reasoning ability, reasoning ability is the machine still needs some guidance. Limit two: the lack of interpretability, artificial intelligence computing is actually very often a black box, can tell you the results, and the results are also right, but it is difficult to explain how to produce this result, because in fact is the dynamic adjustment of the various parameter weights. Limitations three: the lack of robustness.

The projected size of the artificial intelligence market is about 680 billion worldwide and 71 billion in China. In this market, Qin Zhi currently provides services of general computing, training computing and inference computing. Our typical users are mainly schools, such as the University of Macau and the Chinese University of Science and Technology. The second is research institutions, like computing institutes, software institutes, etc. The third line is enterprise users, now the Greater Bay Area in Guangdong's new generation of artificial intelligence development plan, very clearly put forward is the three core development cities, Guangzhou, Shenzhen, Zhuhai, because we are in Zhuhai, but also do one of the three core pivot point.

Linlin Shen: "GAN face changing and application, giving AI applications more imagination



(Pengcheng Scholar, Director of the Institute of Computer Vision, Shenzhen University, Linlin Shen)


Pengcheng scholar, Linlin Shen, director of the Institute of Computer Vision at Shenzhen University, shares with you a hot direction in face, also called deep adversarial networks. What is a generative adversarial network GAN? Linlin Shen explained that the main structure of GAN consists of a generator G (Generator) and a discriminator D (Discriminator).

The generator generates the dummy face through the neural network. The discriminator has to discriminate whether the face is real or fake, the generator wants to generate the face as real as possible, and the discriminator tries to identify the fake generated face, these two parties play each other, there is a game theory, and later the face generated by the generator becomes more and more realistic, and finally fooled the discriminator, and the discriminator can not tell whether the generated face is real or fake, which is the process of mutual confrontation, and finally these two networks are established. We call it the generative adversarial network.




In Mr. Linlin Shen's sharing, he focused on face editing based on Generative Adversarial Network (GAN), introducing several classical networks in the development history of GAN technology such as GAN, pixel2pixel, StarGAN and StyleGAN; then introducing our research work on transforming face attributes based on the latest StyleGAN, and the technique of accurately transforming only the target region of the face without changing other regions; and finally introducing how to use these face transformation techniques to improve the performance of face recognition in the case of unbalanced training data.

Yuhang Jia: "Scenario-based AI Data for Smarter Artificial Intelligence


(Jia Yuhang, General Manager of Beijing Cloud Measurement Information Technology Co.


Jia Yuhang, General Manager of Beijing Cloud Test Information Technology Co., Ltd. shared the theme of "Scenario-based AI data, making AI smarter". Jia Yuhang said: "The 3 cornerstones of AI are computing power, algorithm and data. So what exactly does the data behind AI mean? The actual is through a series of processes, from production, processing to processing such processes are completed before the data can be used by artificial intelligence for data training.

Along with the application of landing, the complexity of the processing scenarios faced by AI increases, and the richness of industry knowledge increases, so the complexity of the training data required by AI gradually increases, and the industry's requirements for data quality also become higher and higher, with data quality directly affecting the accuracy of algorithms, which in turn directly affects the experience of using products. Therefore, in the process of data service, nothing is more important than highly accurate data labeling after data collection.

First, based on the richness of the scene, the cloud test in the country to establish a number of scene laboratory, based on their own understanding of finance, education, home, security, driving and other fields, the accumulation of a rich scene library, sample library to respond to the actual implementation of different scenarios faced by enterprises corresponding to the needs of landing, build the corresponding scene, 100% restore the actual user use conditions, in order to obtain real data.

Second, in the whole data inference, through the way of self-built data labeling base, and with their own set of corresponding theoretical training system, methodology, etc., theoretical foundation and theoretical knowledge of the relevant industry professionals training, and with the universal full range of labeling tools, can support language, text, image and other different areas of labeling capabilities and methods.

Third, in the face of increasingly complex annotation needs, the establishment of a comprehensive process system and management methodology to ensure the achievement of high quality annotation data accuracy in the face of complex scenarios.

Fan Xutong: "ZTE Netcom, AI Empowered Smart City


(Fan Xutong, Deputy General Manager of Shenzhen ZTE Netcom Technology Co.


To recreate environmental governance with data-based decision-making thinking and build a modern environmental governance system based on technology and innovative models is an important guarantee for building ecological civilization and beautiful China. Fan Xutong, deputy general manager of Shenzhen ZTE Netcom Technology Co., Ltd, shared the theme of "ZTE Netcom, AI Empowered Smart City", discussing how ZTE Netcom uses "AI environmental brain" to help urban environmental governance and ecological civilization construction.

ZTE Netcom is not a professional AI company, it is actually a comprehensive solution provider in the industry. ZTE Netcom's strength is its deep understanding of the industry and the huge amount of data obtained from the implementation of industry projects. ZTE Netcom mainly has in-depth accumulation in smart city, environmental protection and medical industries.

Among them, ZTE Netcom in the field of environmental protection artificial intelligence applications can be divided into four parts. First, data analysis and research, this piece is mainly based on big data and deep learning to research; second, pollution monitoring, pollution monitoring is mainly focused on security equipment manufacturers, such as surveillance cameras built-in algorithms; third, business environment optimization such as key polluting enterprises, waste tracking, environmental protection business link optimization; fourth, intelligent governance, intelligent governance is mainly intelligent environmental protection equipment.

NetInfo AI environmental brain's three main application directions are air pollution source detection, water environment detection, abnormal law mining / pollutant traceability analysis. Up to now Netsense environmental AI technology has been applied on some scale in Jinan, Jining, Anyang, Fenyang, Puyang, Lianyungang, Chengdu, Shenzhen Longgang, Guangming District, Dongying, Dalian.


ZTE Netcom's "AI + environmental protection" solution truly integrates AI into many business aspects of environmental protection, truly improves business efficiency, makes truly useful environmental AI applications, and is the leader in AI empowerment in the environmental protection field.

Bruce Lau: "Credit 3.0, Reinventing Intelligent Finance


大路网络科技有限公司首席数据官Bruce Lau

(Bruce Lau, Chief Data Officer, Big Road Network Technology Co.


Bruce Lau, Chief Data Officer of Dalu Network Technology Co., Ltd. brought a sharing on "Credit 3.0, Reinventing Intelligent Finance", Bruce Lau said, "The change of enterprise assessment method under artificial intelligence technology, the existing enterprise credit assessment method in the market is too traditional and backward, extremely relying on financial data and collateral, in the absence of information or financial falsification Under the influence of multiple factors such as lack of information or financial falsification, financial institutions cannot correctly assess the credit level of MSMEs and lack methods and means to effectively control the financing risks of MSMEs; at the same time, some enterprises with good actual business conditions do not have high quality credit under the traditional credit assessment system, and their credit value is greatly underestimated! Only by overturning the traditional way of enterprise credit assessment can we fundamentally solve the problem. Credit 3.0 using AI technology can will completely change this status quo and solve the survival problem of SMEs' difficulty in obtaining loans."  

The Global AI Entrepreneur Conference is now held for the fourth time, which aims to build an exchange platform for leading researchers, enterprises, governments, and investment institutions in the field of AI nationwide and globally. The conference focuses on the AI industry, covering the policy direction of AI, the frontier of technology development, and the operation, product development, market application orientation and commercial implementation of enterprises, to create a top AI industry exchange platform that connects governments, academia, enterprises, and investment institutions.