There is a new development in the construction of the national new generation of artificial intelligence open innovation platform for vision computing. At the 10th China International Social Public Security Expo, the AI chip vision computing innovation platform opening plan was officially launched, which is one of the important contents of the platform construction.
In August this year, the Ministry of Science and Technology issued the "Guidelines for the Construction of the National New Generation of Artificial Intelligence Open Innovation Platform", pointing out that further promote the construction of the national new generation of artificial intelligence open innovation platform, and promote the innovation and development of artificial intelligence technology and industry in China. The new generation of artificial intelligence open innovation platform is mainly led by leading enterprises in the artificial intelligence industry technology, and encourages joint construction by scientific research institutes and universities to provide intellectual and technical support.
Subsequently, at the 2019 World Artificial Intelligence Conference held at the end of August, the Ministry of Science and Technology announced the third batch of enterprises of the national new generation of artificial intelligence open innovation platform. Among the list, the vision computing national new generation of artificial intelligence open innovation platform will build an open source open vision computing platform and community, and promote the development of intelligent chips, intelligent algorithms, and intelligent products.
"When developing algorithms and chips, we found many pain points in the industry. The most critical is that the cost of hardware and software docking is very high." Xu Yuan, Vice President of Business Technology at YITU Technology, introduced that people who develop AI algorithms often do not understand chips, so when algorithm developers want to put the algorithm on the chip for use, they need the chip or hardware manufacturer to send someone for training; conversely, chip manufacturers want to optimize for a new algorithm also need to make a special docking, which leads to a long development cycle.
Especially with the emergence of 5G, users' requirements for AI application latency will be higher and higher, and the coupling of AI chips and algorithms is particularly urgent.
"I hope to share the 'pits' stepped on in the entrepreneurial process and growth experience, and good tools, through a platform to share with the industry, to promote the development of the entire industry." Talking about the original intention of building the national new generation of artificial intelligence open innovation platform for vision computing, Xu Yuan said.
The platform focuses on solving the bottleneck problems of hardware and software docking and technical product docking. Through the platform, AI algorithm developers, AI chip developers, and AI product developers are united, and through collaborative development of hardware and software, an integrated module that is easy to use for AI products is created, truly achieving layered decoupling and open interconnection.
How to achieve layered decoupling? The platform proposed the concept of "hardware and software integrated components". In simple terms, with the help of the platform, the hardware and software integration technology at different levels is encapsulated into independent, convenient docking modules. AI algorithm developers, AI chip developers, and AI product developers choose the most suitable ones for their own use, combining flexibly like building blocks, greatly reducing the time and threshold for AI development.
And the AI chip vision computing opening plan launched this time aims to jointly establish visual computing hardware and software collaborative development standards, output "plug and play" integrated intelligent module development kits, establish an open source community, and jointly promote two-way interaction between the industry and academia with universities and research institutions.
Some people say that the development of artificial intelligence needs to cross the "three mountains" of algorithms, computing power, and data. "In fact, algorithms, computing power, and data are the driving force and source of AI development, and the real 'mountain' is industry knowledge." Xu Yuan said, "Our goal is to combine these four conditions comprehensively to get the whole AI wheel turning."
At present, the platform has established a vision computing laboratory with units such as Shanghai Jiao Tong University. Wu Min, Vice President of Business Technology at YITU Technology, said that the next step will be to open the internally incubated version, serve chip developers, algorithm developers, and product developers more automatically and standardized, and build a collaborative innovation ecosystem.
You can copy the link to share to others: //dribblessportsbar.com/node/903
