"Science and Technology Innovation Board Daily" (reporter Hu Jiaming, Beijing) news, "Deep learning framework in the artificial intelligence technology system, in the position of the waist through the upper and lower levels, it is connected to the chip, the upper application." On March 31, at the Baidu AI Open Day, Dr. Ma Yanjun, General Manager of Baidu AI Technology Ecology, talked about his views on the competitive landscape in the field of deep learning, the development and breakthrough of China's self-developed deep learning framework, and the future trend.
Similar to windows in the PC era and IOS and Android in the mobile Internet era, the deep learning framework is the operating system of the intelligent era. Together with chips, it forms the infrastructure of artificial intelligence, and the deep learning framework is no less important than chips. In the 14th Five-Year Plan, the "Deep Learning Framework" was included in the field of "New Generation Artificial Intelligence" and became a cutting-edge innovative technology supported by the state.
Domestic deep learning frameworks face three major difficulties
Ma Yanjun believes that deep learning frameworks are making AI applications simpler. Based on the deep learning framework, enterprises can develop AI applications faster and more conveniently according to the characteristics and scenarios of their own industries, and no longer need to build foundations from 0 to 1, which greatly improves the efficiency and level of industrial intelligence.
Whether from the perspective of AI technology development or industrial application, deep learning frameworks are in a very core position. Since 2013, the global artificial intelligence academia and various research and development entities in the industry have successively opened up their own research and development deep learning frameworks, and built an open platform for artificial intelligence with the framework as the core to promote the establishment of an artificial intelligence industry ecology. The deep learning framework represented by Google's TensorFlow and Facebook's PyTorch two deep learning frameworks started early and developed quickly, occupying a dominant position in the industry.
As early as 2017, the National Development and Reform Commission officially approved the establishment of a national engineering laboratory for deep learning technology and application, and China's deep learning framework gradually broke through from international competition. In 2021, IDC reported that Baidu Feipao, China's first open source and open deep learning platform, has surpassed other international giants in its comprehensive share in China's deep learning market and become the first in China. This makes mainland artificial intelligence technology developers and users do not have to rely on foreign platforms, and can further rely on domestic platforms to cultivate industrial ecology.
However, China's self-developed deep learning framework still has a long way to go to gain a leading position in international competition. Ma Yanjun pointed out that the current development of China's deep learning framework still needs to break through three key points: technical strength, functional experience, and ecological scale.
First of all, in terms of technological innovation, the research and development of deep learning frameworks requires low-level technical talents in the field of artificial intelligence, and the mainland's reserves in this field are still insufficient.
Secondly, in terms of application experience, since China is the country with the most complete industrial chain in the world, the industrial system is complex, and the transformation needs of small and medium-sized enterprises are imminent. However, in the process of applying AI and promoting the intelligent transformation of enterprises, it takes at least 3-6 months for only one technology application, from the laboratory to the industrial landing. A low-barrier-to-entry or even zero-threshold development platform is extremely important.
In terms of developing the application ecology, deep learning is a typical co-creation technology field, and only by building its own ecology can iterate and develop continuously. However, the construction ecosystem cycle is long and the cost is high, and only when the technology and functional experience of the domestic framework are sufficient to meet the needs of developers, will there be an opportunity to cultivate an independent and innovative AI development application ecology.
The deep learning framework may determine the AI industry pattern in the next 5 years
In the field of global deep learning, foreign developers mainly develop, train and deploy artificial intelligence algorithms and models based on foreign deep learning frameworks such as TensorFlow, PyTorch, and MxNet. There is still a certain gap in the deep learning framework developed by Chinese intelligent enterprises in terms of community prosperity and number of developers.
However, China's deep learning framework, represented by Flying Propeller, is developing into an open source open platform that is more suitable for industrial needs and more popular with Chinese developers. On the one hand, China's deep learning framework continues to take root in practical application scenarios, firmly grasps the needs of developers and enterprises to upgrade intelligently, and reduces the application threshold of artificial intelligence technology. On the other hand, China's deep learning framework has been deeply adapted and integrated with more chip manufacturers, forming a soft-hard synergy advantage.
"Chinese enterprises and industries have their own characteristics, for example, in the fields of industry, agriculture, logistics, finance and other fields, the demand for AI technology of Chinese enterprises also has its own uniqueness. If the domestic deep learning framework can meet the needs of China's industry in a large number of functions, and at the same time, it is low threshold and simple and easy to develop, it will have a great opportunity to achieve curve overtaking on the industrial level. Ma Yanjun said.
What cannot be ignored is that China's deep learning framework is still facing difficulties such as complex adaptation and deployment, difficult application development, etc., and the construction of an independent and controllable deep learning and artificial intelligence industry ecosystem is obstructed and long, but it may determine the AI technology pattern and industry level in the next 5 years. Ma Yanjun said: "Although the deep learning framework belongs to the competition of high investment, long cycle and ecological grabbing, it has received strategic support from the state and enterprises, and is the key to opening the next AI era. “