Historically, AI technology has experienced two cold winters, both because of the disconnect between technological development and industrial needs. Since AI technology entered the third wave of development in 2017, all circles of industry and academia are asking the same question: this time, AI technology represented by machine learning can be combined with industrial development and enterprise needs, thereby avoiding another cold winter. coming?

Over the past few years, AI technology has also experienced many controversies. But it can be seen that AI is going further and further in the core direction of integrating into the industry. Enterprises' recognition of intelligence has gradually increased, and the basic capabilities in the field of machine vision and intelligent voice have been implemented in large-scale industries. At the same time, a more critical question has emerged: Is the goal of AI to become a number of basic capabilities that can be invoked by enterprises, or to help enterprises achieve the liberation of productivity from a deep level?

This problem directly determines the upper limit of the value of AI, and also determines the upper limit of the integration of cloud services, AI, and enterprises. The core of this question is whether AI can truly be integrated into the core production system of an enterprise.

AI technology is integrated into the core of the industry, which can be said to be an intelligent deep-sea area, which is still in the stage of exploration and experimentation. On this road, HUAWEI CLOUD is one of the farthest cloud computing service providers. From the perspective of market data, HUAWEI CLOUD ranks first in China's machine learning public cloud service market share in the latest "2021H2 China AI Cloud Service Market Research Report" released by IDC. This shows that enterprises are more accepting of the AI ​​development platform provided by HUAWEI CLOUD to complete complex and in-depth intelligent construction. The key point is the AI ​​development production line ModelArts provided by HUAWEI CLOUD.

Let us see how HUAWEI CLOUD and ModelArts complete this AI deep dive from three aspects: market, technology, and ecology.

AI deep sea, located in the core of the industry

There are two main ways for enterprises to apply AI capabilities: standardized invocation of AI capabilities and customized AI development. Obviously, standardized AI capabilities, such as machine vision, speech recognition, NLP, etc., can only solve specific problems. In order to extract value from AI technology in an in-depth and normalized manner, companies inevitably need to carry out a lot of AI development work. The closer the AI ​​development to the core production scenarios of the enterprise and the activation of the productivity of the main road, the more value it can bring to the enterprise.

It may be said that AI capability invocation is the shallow water area of ​​AI; AI development is the central water area; AI development and deployment that go deep into the core of the industry is the real AI deep sea.

To capture value in the deep sea, companies generally face two problems:

First, there are too many steps in AI development, and the process is too complicated, involving steps from data collection and processing, to AI training, reasoning, and application. And a large number of new technologies emerge at each step of the meeting, making it impossible for companies to start. Development barriers between each link can also cause a lot of trouble.

Second, there are too many factors involved in AI development. Enterprises need abundant computing power, effective development tools, experienced AI development talents and other conditions. The absence or weakness of any one factor will cause an overall AI development dilemma.

The existence of these two difficulties makes AI development close to the core of the industry into a "cask dilemma". Even if an enterprise already has many capabilities and conditions, there may still be problems such as a surge in AI development costs, slow launch time, and poor business compatibility due to one or several weaknesses.

Today, it is not the optimal solution for enterprises to build their own development platforms and develop AI applications. Finding a suitable professional development platform is crucial. Machine learning development through public cloud is becoming the choice of more and more enterprises.

The Machine Learning Public Cloud: A Certified Deep Dive

Relying on the cloud base for machine learning development, relatively speaking, it can plan resources more reasonably, flexibly access scenario-based capabilities, and deploy computing power on demand, which is being accepted by more and more enterprises. Judging from the current market space, the rising speed of AI public cloud is much faster than that of other AI software service models.

AI cloud services are divided into several types. Among them, machine learning cloud services are the one that best meets the needs of enterprises for customization and platformization, and are also the one that tests the comprehensive AI capabilities and development tools of cloud service providers. It can be said that the machine learning public cloud is one of the most ideal channels for the intelligent deep-sea area of ​​​​the industry, which has been deeply recognized by enterprises and industries.

In this channel, HUAWEI CLOUD has taken the lead. According to the latest "2021H2 China AI Cloud Service Market Research Report" released by IDC, in the second half of 2021, Huawei Cloud ranked first in China's machine learning public cloud service market share. So far, ModelArts, the HUAWEI CLOUD AI development production line, has topped the market for four consecutive times.

When looking at the composition of HUAWEI CLOUD in the machine learning public cloud industry, two notable features can be found: one is rapid growth, and the market share has soared; The demand for transformation is placed on HUAWEI CLOUD.

In this key channel, how does HUAWEI CLOUD achieve market leadership? The answer lies in two aspects, technology and ecology.

The deep sea needs "all-round submarines": an interlocking AI production line

As mentioned earlier, in enterprise customized AI development, the number of steps and complexity is the first problem. Especially for large and medium-sized enterprise customers represented by government and enterprises, the business process itself is very complex, and AI capabilities need to be embedded in the industrial system.

In order to help enterprises achieve this goal, the breakthrough solution is to upgrade the technology of the development platform, so that the development can move towards the automation and intelligence of the whole process. As the first batch of leading AI development platforms certified by the China Academy of Information and Communications Technology, ModelArts has always aimed at lowering the development threshold and opening up the development process. ModelArts, which debuted in 2018, focuses on intelligent data annotation and automatic model training. It brings a one-stop development experience to enterprise developers. Over the past few years, ModelArts, which has been continuously upgraded, has continuously enriched product functions in all aspects of the AI ​​lifecycle, including data processing, algorithm development, model training, model management and deployment, and finally built an "AI development production line".

In the data processing stage, ModelArts provides massive data preprocessing and intelligent labeling capabilities, saving enterprises a lot of repetitive labor and development time; in the training stage, ModelArts can support large-scale distributed training, which is closer to the development needs of industrial-level AI, and can It realizes the automatic generation and high interpretability of models, which greatly reduces the development threshold; in the field of deployment, ModelArts can realize on-demand deployment of all scenarios to meet the complex scenario requirements of enterprise application AI.

In this way, all aspects of AI development can be opened up by ModelArts, and the full-cycle workflow management of AI can be completed. From the realization of the whole process to open up, low threshold, low-cost AI development. Not long ago, HUAWEI CLOUD AI development production line ModelArts also added a new member: HUAWEI CLOUD ModelBox AI application development framework, which provides users with the generation, evaluation, inference and deployment capabilities required for application development, helps developers shield the underlying software and hardware differences, and achieves One-time development and full-scenario deployment of AI applications.

If you want to dive into the complex, changeable, and high-pressure deep sea, a submarine must have every ability. To make AI go to the core of enterprise production, what is needed is not one kind of technology or one product, but an interlocking AI production line formed by the superposition of countless technologies.

Gathering Voyages with Ecology: Building an AI Deep Sea Exploration Cluster

Another difficulty in enterprise AI development is that there are many influencing factors, and the support and help that enterprises need are also diverse. This cannot be solved only by technology, but by building a comprehensive ecological environment, so that every need of enterprises can find answers and help in the ecology.

In order to achieve this goal, HUAWEI CLOUD has built a rich and three-dimensional ecosystem around AI development. Together with partners, users, and developers, we have reached the deep-sea area of ​​industrial intelligence.

At this stage, the AI ​​ecosystem built by HUAWEI CLOUD can be understood as two parts: vertical and horizontal. Horizontally, HUAWEI CLOUD AI Gallery is a developer community built on the basis of HUAWEI CLOUD AI development production line ModelArts. It not only meets the needs of developers, but also meets the sharing and transaction needs of various roles in the AI ​​ecosystem, thereby realizing the integration of the AI ​​ecosystem. All links are opened to further reduce the development threshold.

From a vertical perspective, in order to deeply meet the AI ​​development needs of enterprises and achieve accurate docking of customized AI capabilities, HUAWEI CLOUD has released the Ecological Partner Program D-PLAN. This plan, by building an AI ecosystem with partners, will fully open up the blocking points and difficulties in AI development, and accelerate the implementation of AI applications through side-by-side collaboration.

Cao Cao Travel is a travel platform that we often use. As a representative platform for its own vehicles, Cao Cao Travel has a strong demand for precise vehicle scheduling. Using AI technology to optimize the vehicle configuration plan, allowing drivers to predict areas with a large number of passenger orders and no traffic jams, is obviously the core production capacity of the travel platform. However, standardized AI capabilities cannot meet the needs of accurately predicting travel supply and demand. Therefore, based on HUAWEI CLOUD ModelArts, with the support of HUAWEI CLOUD D-PLAN, Cao Cao Travel and HUAWEI CLOUD jointly created an intelligent scheduling solution. This plan simultaneously exerts efforts in order forecasting and vehicle scheduling, optimizing the platform's supply and demand control capabilities, and promoting revenue growth.

From the final result, HUAWEI CLOUD's AI solution has brought a shorter development cycle to Cao Cao Mobility, greatly reduced development costs compared with the self-research path, and shortened the delivery cycle of order forecasting capabilities and vehicle scheduling capabilities. 30%. And this cooperation fully activates the advantages of cloud services, and makes use of the computing power of HUAWEI CLOUD to improve model training and deployment efficiency by more than 20%. In the end, Cao Cao's drivers can use the "eyes of AI" to foresee where there will be orders, and effectively reduce the time for drivers to cruise around empty cars.

At present, HUAWEI CLOUD AI Gallery and their AI ecological plan D-PLAN have gathered the strength of more than 50,000 AI-related assets, more than 70 universities and more than 200 doctors, covering six major industries and 25 sub-scenarios. For the objective situation of enterprise AI development volume, complex requirements, and high degree of customization, HUAWEI CLOUD D-PLAN empowers it in a more reasonable and direct way. Help enterprises quickly obtain value returns from AI capabilities.

Facing the challenges and responsibilities of the AI ​​deep-sea area, HUAWEI CLOUD chose the high-value track of machine learning public cloud; nurtured the AI ​​development production line ModelArts, an all-around submarine; and through the development of community and cooperation plans, created an ecology of group ship exploration voyage.

The convergence of forces from several aspects finally allowed HUAWEI CLOUD to be recognized by the AI ​​market. Under the deep sea, is the treasure of the intelligent world.

Leave a Reply

Your email address will not be published.