Committed to building an open AI ecosystem! Exclusive interview with Mr. Li Ying, Vice President of Intel Corporation

Although Intel has a vision to realize ubiquitous AI, in addition to hardware preparation, developers of AI applications must also have suitable development tools, an open ecosystem, and reliable technical support. , Only in a safe environment can we quickly and massively develop AI applications that users need. So what kind of support does Intel have for developers? To this end, we conducted an exclusive interview with Mr. Li Ying, Vice President of Intel Corporation and General Manager of Intel China Software Ecosystem Division.

Mr. Li Ying, Vice President of Intel Corporation and General Manager of Intel China Software Ecosystem Division

OpenVINO will become an incubator for client and edge AI applications

First of all, Mr. Li Ying introduced us to the OpenVINO open source tool software launched for client and edge computing platforms. Since many users’ AI applications are not in data centers, they do not need to use a large model training environment. Most of the time, AI functions are brought to the edge or client such as production, retail, factories, and homes to popularize AI functions. Therefore, Intel launched the OpenVINO development tool, which is also the greatest value of OpenVINO. The software provides many ready-made APIs, models, and architectures. Developers do not need to know how these models need to be trained or how to model, they can use them directly, which can truly bring AI capabilities to application scenarios.

Currently in China, OpenVINO has been widely used, and Intel holds a special OpenVINO summit every year, including OpenVINO workshops and OpenVINO competitions. In China's OpenVINO competition, we can see a variety of different artificial intelligence application scenarios. For example, there was a young man who had just graduated from college who left a deep impression on Mr. Li Ying. During his internship, he discovered that there were many safety hazards in the production line. For example, some people did not wear gloves, which caused many safety problems. Therefore, he designed a simple function in the game that uses visual detection to identify whether each worker is wearing gloves, and triggers an alarm if no gloves are worn. Therefore, even a newly graduated college student can quickly design artificial intelligence applications through OpenVINO. This is due to Intel OpenVINO integrating a variety of different models into its library, including models developed by third parties, making it easy for users to use. With the popularization of artificial intelligence computers, users will find that OpenVINO has more usage scenarios, including various visual and text extraction models, etc.

OneAPI can adapt to a variety of hardware, so it has an expert center and professional team in China

In key aspects of building AI applications, Intel provides OneAPI development tools, which can support various high-performance hardware, including CPUs, GPUs and FPGAs. In addition to trying its best to adapt to various major hardware products, Intel has also established multiple expert centers around the world, including the Chinese Academy of Sciences and Peking University in China. This is because Intel doesn't know how many different hardware devices there are around the world, so an important role of the Expert Center is to share OneAPI best practices through the network. In addition, the expert center is also responsible for adapting various hardware devices to realize the overall concept of OneAPI

The Chinese Academy of Sciences and Peking University have fully adapted OneAPI to some domestic hardware. Therefore, if domestic hardware manufacturers and software manufacturers want to adapt through the OneAPI network, they can do so in two ways: one is through the expert center, and the other is through Intel China's dedicated OneAPI team. In addition, various domestic standards organizations are also designing some standards for heterogeneous computing, and Intel is also actively participating in discussions on these standards. Intel hopes to promote the establishment of China's entire heterogeneous computing standard through the OneAPI concept and technology

Build Intel Developer Cloud Platform and cooperate with many manufacturers to accelerate innovation

In order to make it easier for developers to access all kinds of new hardware, including Intel's GPUs, CPUs, and Intel's various AI architectures, and to more easily deploy and optimize their models, Intel's most important step is to launch Intel Developer Cloud Platform. Mr. Li Ying said that the developer cloud platform is essentially a cloud service that allows developers to use Intel's latest applications and the latest software architecture, and integrates many functions of OpenVINO and OneAPI into the developer cloud platform. In addition, in order to help developers develop applications that can meet various security requirements, Intel is not only cooperating with global vendors such as Red Hat, Canonical, and SUSE, but also with local open source vendors in China, including OpenEuler and OpenAnolis OS. It is to use Intel's innovations in hardware and software and the innovations of other manufacturers through Linux distributions to enable developers to use them faster, empower developers, and empower partners to accelerate innovation.

Establishment of Intel China Open Source Technology Committee

There are a large number of open source projects in China, so the last and most important step is for Intel to establish the Intel China Open Source Technology Committee. The members of this committee come from various industries in Intel China. The main purpose of this committee is to establish regular and close cooperation with China's open source organizations, apply global advanced technologies to China more quickly, and find unique and innovative projects in China. Through the joint efforts of Intel and its partners, these open source projects can develop faster and help China's open source organizations achieve better and faster growth

At the end, what needs to be rewritten is: write at the end

Currently, there are two main types of AI ecosystem: closed and open. On the one hand, since the closed ecosystem does not need to consider the needs of adapting to different hardware and partners, its software and hardware products generally develop very quickly. Many products It has become the core of AI computing. However, the problem is that these products are susceptible to restrictions in various aspects, and it is difficult for the demand side to obtain core technologies from them, so they must rely on external supply. Therefore, for users, a safer way is to pay attention to and develop the open AI ecosystem at the same time, and cooperate with manufacturers like Intel who are committed to creating an open ecosystem, so that developers can develop more and better AI products. Intel products It can also be adopted by more users, allowing both parties to achieve a win-win situation.

The above is the detailed content of Committed to building an open AI ecosystem! Exclusive interview with Mr. Li Ying, Vice President of Intel Corporation. For more information, please follow other related articles on the PHP Chinese website!