Shulou(Shulou.com)06/02 Report--

Boundless, angry sea, a submarine, 135 sailors, 180-day voyage, how to make every sailor an excellent leader?

That sounds a little unlikely, said L. L., commander of the submarine Santa Fe. David Marquette did it. He used a new model-"leader + leader", to train each sailor into an independent leader and master the ship's "absolute sea domination" in rough seas.

This story immediately reminds me of using the talent training model of "master to master" to promote the industrial AI of Whampoa College.

The "Huangpu College" jointly built by Baidu and the National Engineering Laboratory of Deep Learning Technology and Application is mainly aimed at the core technical personnel of the enterprise, provide about half a year of deep learning hard core training, and train the "chief AI architect" for industrial ports.

How to turn enterprise talents from various industries into chief architects who can lead the industry's AI warships to meet the wave of industrial intelligence is no easier than the rough seas of the physical world.

In order to understand the thoughts of these "AI Mariners", I specially chatted with many students from different industries at the recent opening ceremony of the third phase of Huangpu College.

From "Magic" to "Application": industrial AI in Deep Water area

On June 6, Huangpu College in Beijing, this is the first offline activity that Xiao Wang, a technical engineer of an aircraft manufacturing enterprise, participated in after the epidemic.

Unlike the well-prepared students who talk eloquently about the application of artificial intelligence in their own enterprises or industries, Xiao Wang's aircraft manufacturing industry is an area with high requirements for the reliability of new technology. the unexplainability of deep learning determines that its current application is still relatively limited. This put a little more pressure on him.

When sitting around, Xiao Wang also faced a lot of "concern" from his classmates: what AI abilities did you use? Dare the plane use AI autopilot? What is the topic of your application?

Of course, Xiao Wang is not alone.

At the scene of the third phase of Huangpu College, there are also many students from oil and heavy industry, industrial power, rail transit, auto finance and other fields.

This coincides with the situation of stronger participation in the Internet digital industry during the recruitment of the first phase of Huangpu College in 2019, showing two different painting styles, and also coincides with the current trend of rapid industrialization and commercialization of AI technology in the wave of new infrastructure.

In this year when AI industrialization and intelligence are the key words for enterprise growth, "Xiao Wang" at the scene of the third phase of Huangpu College is like a traveler who has been hiking in the desert for a long time, and has finally completed a meeting with his companions in the oasis.

(photo: group photo of the students in the third phase of Baidu Huangpu College)

These technical elites from various industries who are eager for AI will complete rigorous in-depth study and training at Whampoa College, and finally graduate successfully through the submission of AI actual combat report and defense. The third phase of students need to complete three hours of high-information real-time study online every week. During this period, group discussion, homework completion and TA communication are also essential.

You may wonder why the course setting of Whampoa College is so different from the ordinary workplace training of taking online classes, handing in topics, group chatting and love.

The core reason is that after the cold ploughing and summer work of AI industrialization, Baidu deeply understands that it is not so easy to plant AI in the industrial soil and harvest the fruits of intelligence.

On the one hand, countless enterprises, large and small, are embracing artificial intelligence.

From worrying about being robbed of jobs and dominating human beings when I first saw AI, I am used to talking to smart customer service, taking photos with AI, paying for facial scans and traveling today. Artificial intelligence has long been a necessity of human life style.

In the industry, whether it is looking for new opportunities for growth after the epidemic, or improving efficiency and reducing costs to maintain survival, artificial intelligence has long become an artifact that can bring real value to the majority of enterprises.

This may also explain why in the third phase of Huangpu College, in addition to more digital enterprises such as the Internet and third-party technology integrators, more and more traditional industries are participating in in-depth learning and discussion.

At the same time, if new technologies want to quickly enter the ports of various industries, it is obviously not enough to rely solely on enthusiasm and intuition to "dream-chasing AI circles".

The Chinese resources of the mainstream overseas framework are relatively poor, and the training is very difficult; the primary machine learning model can not support the high-precision needs of the industry, and AI can only stay at the ideal level; the cost of recruiting algorithmic talents is getting higher and higher because of the various technical documents; buying solutions directly from third-party service providers may not accurately match the actual needs. These are the puzzles and problems that many enterprise managers and technical teams will face when introducing artificial intelligence into specific business.

In the final analysis, as an efficiency tool, the application of artificial intelligence also needs to be combed and changed from cognition, talent, process and business within the enterprise, so as to find the appropriate platform and ecology in order to become the master of technology and realize the entry and landing of AI.

This is also the original purpose of the establishment of Whampoa College-- through the collaborative participation of platform, enterprise, academic and other forces, the efficient cooperation between technology and industry, the exchange of needed goods, and the training of "Chief AI Architects" who not only understand application scenarios, but also understand AI technology, and can apply AI technology to solve practical problems in the scene.

At a time when AI is spreading thin to the whole society, and when intelligence has become the main theme of various industries in China, the third phase of Whampoa College has also arrived as promised, and will help a new group of students "turn AI into application" in the next three months.

Mid-stream hits the water, waves suppress the flying boat: Baidu's intelligent thoroughfare carved with three heavy equipment

Here, you may have new doubts, why Huangpu College, why Baidu?

When AI has become the main theme of the transformation of Chinese enterprises, supporting developers and promoting industrialization has become the due meaning of science and technology enterprises. Various competitions, communities, university industry and research programs have emerged one after another, and why Whampoa College, which has focused on the landing of industrial intelligence from the very beginning, can stand out among so many talent training programs.

In my opinion, it is because Baidu has carved out a thoroughfare that allows AI technology to flood through three major heavy weapons.

1. AI, the industry of flying close to the ground

The distance between artificial intelligence and industry has never been so close. How to give algorithms, tools and environment to industry developers to give full play to their talents requires not only the strategic position of AI technology research and development, but also the accurate pulse of real business scenarios.

Baidu AI is characterized by a wide range of business scenarios, accumulating huge amounts of data in business scenarios such as information flow, search, map and O2O, and creating a high-precision AI model with R & D strength and technical layout in the fields of natural language processing, machine vision, knowledge graph and so on.

These modular tools, which can be called by different industries and different needs, are released through the Baidu brain AI open platform, which is undoubtedly a sweet spring for the anxious industrial land eager to be washed away by the technological water flow.

It is precisely because of an in-depth understanding of the pain points of the industrial application of AI, Baidu is well aware that entry-level education and universal training are a drop in the bucket for the industrial AI, and that it is necessary to enter the actual combat and hard core output in order to drive the AI of various industries to blossom.

(photo: Wu Tian, vice president of Baidu Group, deputy director of the National Engineering Laboratory of Deep Learning Technology and applications, and dean of Huangpu College)

Therefore, in the application of the third phase of Huangpu College students, we need to have a certain plan for the problems faced by the industry, the potential landing of AI, and the direction of industrial practice that I am interested in.

In the exchange with the students, an intelligent customer service product director from the communications industry told me that since I contacted Baidu in 2017, I began to access Baidu's enterprise-level voice recognition model. it is applied in the whole business process of the enterprise intelligent customer service before, during and after the conversation. With the help of AI ability, real-time and accurate speech recognition into text, to help ordinary users bid farewell to the menu-style tedious service, to achieve one-click voice to solve specific problems. At the same time, it can fully release the potential of voice big data, and further help enterprises improve quality and efficiency in customer service dialogue, such as comprehensive quality inspection, service value mining and other scenarios.

With its rich technical modules, open industrial energy release and approachable service ecology, Whampoa College, which is based on Baidu AI technology, is more suitable to bear the important task of training industrial AI talents.

(photo: the third phase of Baidu Huangpu College)

two。 Self-proof of the ability of flying paddle platform

A cadet who did not want to disclose the unit told me that his military field has a very strong demand for the localization of intelligent technology. Although he has a very high accumulation of technology in the cutting-edge areas such as artificial intelligence, image processing, and man-machine collaboration, he is still willing to participate in the training of Huangpu College, that is, he wants to learn more about PaddlePaddle, an open source and open deep learning platform.

In fact, the in-depth understanding of the flying oar platform is almost one of the core demands of all students.

Readers familiar with the field of artificial intelligence know that Flying Propeller is the most representative Chinese deep learning framework and industrial platform at present, which is based on Baidu's years of deep learning technology research and business applications. the integration of deep learning framework, basic model base, end-to-end development kit, tool components and service platform can help developers lower the technical threshold. It is easier to complete the integration of deep learning and their own industry.

A student from Lianxin Medical told me that after the completion of the first and second phases of training at Huangpu College, their company has begun to rely on flying oars to develop related medical imaging functions, such as image semantic segmentation and other techniques to improve the accuracy of recognition and assist doctors to complete clinical diagnosis.

After the epidemic, the demand for digitization and intelligence in hospitals has soared, and flying oars have become more friendly to Chinese developers, all of which have made their company more motivated to continue to participate in the Huangpu College project. More in-depth with Baidu to build core technological advantages based on flying oars.

In addition to the independent development of the Chinese framework and the open source of Baidu's deep learning algorithm, the full autonomy of the flying oar tool is also more friendly to traditional enterprises that want to taste AI.

Because even if the call interface such as SDK is provided, for some enterprises that do not have the development capability, they still need to find a third-party service provider to provide packaging solutions. But how does the standardization scheme adapt to the specific conditions and data of the enterprise itself? through the enterprise version of EasyDL zero-threshold AI development platform, flying oars provide high-precision AI model training services for engineers and small and medium-sized enterprises based on zero algorithm, and one-stop AI development process capabilities such as data management and tagging, model training, model deployment and so on.

Aiming at the AI processing, training, analysis, reasoning and other AI capabilities needed by large enterprises, the full-function AI development platform BML, which has been tested by Baidu's internal search and recommendation applications, has released applications with comprehensive functions, flexible customization and deep integration. How hard is the core exactly? A model that requires hundreds of T samples to train can be done in a few hours on BML.

It is conceivable that when the students of the third phase of Huangpu College sail to various industrial ports such as industry, agriculture, energy, communications, aerospace, and so on, the "new infrastructure" aircraft carrier driven by artificial intelligence will also raise its sails at a higher height than expected and set sail for the future.

3. Baidu AI talent team

Artificial intelligence is a competition for the mental power of talents. This is true for industrial intelligence, especially for high-tech platform enterprises.

To take the simplest example, no climber with the best physical fitness can safely conquer Mount Everest without the help of a guide.

For the platform, not only output technical capabilities, open AI resources, need the strength of countless scientists and engineers. One of the core "weapons" of Whampoa College is the large-scale super talent system.

Judging from the faculty allocation of the third phase of Whampoa College, it not only continues the "super lineup" of the first two periods, which is taught by a number of Baidu heavyweight scientists and AI architects, but also specially configures an expert group to guide the whole process, with excellent PM and other employees from Baidu as teaching assistants to provide comprehensive support before, during and after the courses such as project construction, process running, adjustment and optimization, and defense presentation.

Taking the graduation report as an example, students will take the specific needs of enterprise application of AI as the core in the learning process, and find and explore solutions to practical problems under the guidance of Baidu senior engineers. As a result, this group of technical talents are like climbers assisted by professional guides, and every step of the climb is rooted in the solid soil of the industry.

The export of human resources of Whampoa College, on the one hand, benefits from Baidu's "engineer culture" of respecting technology, and absorbs a group of top international academic talents; it also works with Baidu to cultivate AI for a long time, through continuous internal application and cultural construction, so that all staff can participate in the support of industrial talents such as Huangpu College and provide strong backup.

In addition to Baidu's internal scientists and senior engineers personally give lectures and guidance, Flying Propeller also gives more developers the opportunity to touch AI through online education.

In the past six months alone, the number of courses studied by flying oars has exceeded 820000, an increase of 16 times.

The enterprise students who graduated from Huangpu College in the previous two periods have helped large and medium-sized enterprises in finance, telecommunications, energy, transportation, Internet and other fields to promote the intelligent upgrading of related industries. Now, the third phase of the students will also carry the oars of AI to the vast sea of digitalization and intelligence of the new infrastructure.

Baidu, which takes the initiative to shoulder the output of AI talents, is not only within the reach of the ability, the vision of the industrial AI, but also the goal of the revival of China under the wave of intelligence.

(photo: the scene of the opening ceremony of the third phase of Baidu Huangpu College)

Today, in the field of AI industrialization, not only the intelligence upgrading of thousands of industries will be triggered by actual combat talents, but also the secret warfare and self-proof of talents between technology platforms.

When we set sail in the third phase of Huangpu College, we were able to see a group of new captains and technical leaders continue to grow, and in the future they will become industrial AI leaders who will fully coruscate the vitality of intelligent technology. In the sea area of the new infrastructure tide, the mid-stream hits the water, and the waves suppress the flying boat.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.