SHENZHEN, China, April 18, 2018 /PRNewswire/ -- On April 18th 2018, the Global AI Application Innovation Summit was held in Wuzhou Guest House, Shenzhen. The summit was hosted by the Shenzhen Science and Technology Association, People's Government of Futian District, China Service Alliance for Integration of Informationization and Industrialization, and National Research Center for the Development of Industrial Information Security; undertaken by Corerain AI Application Innovation Research Institute, Shenzhen Original-Creativity Offshore Innovation Center, Shenzhen High-caliber Talents Association, and Shenzhen Innovation Presidents Club; and co-organized by Corerain Technologies. On this grand gathering, international AI experts and leaders from the Shenzhen government, the Futian government, and the China Service Alliance for Integration of Informationization and Industrialization inaugurated the Corerain AI Application Innovation Research Institute together. The Corerain AI development platform fills the void in the AI chip industry for edge devices and boosts the innovation and development of the industry as a whole.
The Unveiling of Corerain AI Application Innovation Research Institute Raises Production and Research of AI Chips in China to a Higher Level
AI development needs support from platform technologies. By constructing an open platform and reducing the technical barriers to entry, more companies and developers will jointly improve the AI industrial chain and apply the technologies to various domains.
Until now, open AI platforms for some specific domains have been launched successively, but none of them has complete low-level integrated circuit chip support. In the upstream of the AI industrial chain, integrated circuit chips are the basis of all the top-level applications. The core technologies of AI chips are of great strategic significance to the AI industry.
At this summit, the unveiled Corerain AI Application Innovation Research Institute filled the void of AI industry in this area. The establishment of this institute was initiated by Prof. Wayne Luk, Fellow of the Royal Academy of Engineering and Co-founder of Corerain Technologies. Based on the state-of-the-art technologies of AI chips, this research institute aims to set up the first open platform which lowers the development barrier of customizable hardware-based AI applications in various domains. The Institute is positioned to establish an artificial intelligence industrialization technology platform to support the latest AI technologies to virtually, quickly and effectively land in various domains, and to fill the platform support voids of the underlying hardware. The technology platform will work in synergy with open application platforms, such as automatic driving, smart city, medical imaging, and intelligent voice, to provide fundamental computing resources. It will accelerate the development of AI ecosystems with domestic independent intellectual properties, and integrate the resources from the top universities and leading AI industries around the globe. The ecosystem also utilizes industrial funding to establish an integrated platform for research, production, and financing.
Corerain AI Development Platform Released, Embedding Intelligent Brains for IoT Application Devices
As well as promoting the research institution, Corerain Technology released an AI platform-level product at the summit - the Corerain AI development platform. This complete AI ecosystem is based on the in-house-developed low level chip architecture and relies on an open development platform. It provides rich computing resources for all artificial intelligence applications. The development platform can automatically generate the DFG of the AI algorithms for applications in the IoT edge devices. It also generates the configurations for customizing the AI hardware. This will really make the edge devices "intelligent" and provide the critical features to facilitate growing top-level applications.
Corerain Technology has released the world's first AI development platform with both the underlying data stream chip architecture and the core technologies of top-level application development. The platform provides an intuitive path from data to solutions.
With a series of advancements in the field of deep learning, more and more domains have begun to use artificial intelligence algorithms based on deep learning to solve practical problems and have achieved very good results. In response to this trend, many companies and research institutes have also launched their own deep learning development platforms. At present, the mainstream AI platform focuses on solutions in the cloud or on the server nodes. Platform support for edge devices is still at a blank stage. With the continuous development of the IoT, the demand for a chip-level development platform for edge devices is imminent. The Corerain AI application development platform has emerged as the times require.
The Corerain AI development platform focuses on artificial intelligence chips and can support full automation from data annotation, model training, and DFG optimization to hardware compilation and board testing. The platform requires users to provide only data labels to automate and customize the AI solutions in specific domains for the edge devices. The entire process does not require any low level hardware expertise or any code writing. This greatly reduces the utilization and adaptation barriers. The application domains and industries which require artificial intelligence development can easily customize their own solutions based on the Corerain AI development platform and use the AI-enabled devices quickly and efficiently.
At the same time, the platform incorporates the Corerain innovated hardware optimization for deep learning algorithms. It supports and accelerates most existing deep learning networks. The Corerain AI development platform encourages and supports AI chip-based solutions in areas such as smart monitoring, smart production, and the Internet of things (IoT). The rapid deployment and adaptation accelerating the burgeoning of AI technologies in edge computing.
Currently, based on the Corerain AI development platform, the prototypes of the "Nebula" and "Rainman" artificial intelligence chip architecture and development platform have been completed. The fixed-point computing architecture supports all deep learning algorithms from the Tensorflow platform. The large-scale deep learning network runs at high speed in a low-power hardware environment. The custom computing architecture provides industry-leading performance. The runtime-adaptable tool chain is extremely versatile. The high-level compilation algorithm also ensures rapid progress from customization to application deployment. These implementations have been adapted in various domains such as aerospace, industrial monitoring, smart city, education and research and development. Corerain Technology has won support from the National High-tech Industry Fund and the EU high-tech industrialization project.
World-renowned experts share frontier breakthroughs in the AI field at the 2018 Global Artificial Intelligence Application Innovation Summit
The summit brought together leaders of government, top international academics in the field of artificial intelligence, top global technology companies, internet giants, industry leaders and lead investors to discuss the development direction of artificial intelligence in commercial development and scientific research. The summit consisted of three parts: government speech, keynote speech and industrial forum. At the meeting, several famous AI scientists from various countries shared the latest progress and application directions in their respective fields.
Dr. Zhou Jian, Director of the Center for Information Research and Promotion of the National Industrial Information Security Development Research Center, shared that the world is undergoing an accelerated transition from an industrial economy to a digital economy. Industry integration, fusion and innovation are also being carried out simultaneously. Under this premise, machine intelligence and human are the core. The development will help promote industrial upgrading, promote open cooperation and collaborative development of industrial resources, and form an artificial intelligence industry ecosystem.
Professor Wayne Luk, international leader on custom computing, shared how to build a development platform for automation of artificial intelligence applications from the lowest level of chip technologies, and then form a complete ecosystem based on this automation platform. He also presented the future path and several real world examples of AI applications in domains such as smart manufacturing and smart finance.
Professor Steve Furber, an international expert in the field of brain-inspired chips from the University of Manchester, introduced the origins and latest developments of brain-inspired computing. The idea of brain-inspired computing started 200 years ago, and it evolved from the first computer more than 60 years ago to the current super-large scale parallel computing chips. It has significantly improved performance. Professor Furber shone a spotlight on the human brain project and the SpiNNaker brain-inspired computing machine developed by their research lab. The machine contains 500,000 computing cores. When compared with the current implementation of traditional neural networks, it has outstanding advantages in terms of performance and energy consumption, making it a leading brain-inspired computing platform.
MIT Professor Arvind, an international recognized figure in field of chip verification, described how to use flash memory technology and hardware accelerators to build large data graph computing systems. The advantages of flash memory over traditional hard drives are the low price and low power consumption. The GraFBoost graph computing system he built is based on external flash storage. By improving the algorithm and embedding the FPGA hardware accelerators in the storage system, an ordinary PC (with the GraBoost system) can achieve server-class performance, thus providing a very low cost and low power solution for large-scale graph computing.
Dr. Derek Chiou from Microsoft shared how the company has integrated FPGA accelerators in its cloud system for infrastructure and deep neural network acceleration, highlighting the AccelNet for accelerating software-defined networks, and the BrainWave stack for DNN acceleration in Bing. By utilizing FPGA's low latency and highly flexible features, it enables higher-performance networks and accelerates deep learning. Under tight requirements of efficiency and cost, it fully meets the different needs from networks and search functions.
Professor Kunle Olukotun, an expert on multi-core computing chips, introduced the DAWN project at Stanford University. This includes collections of algorithms, methodologies and tools for developing machine learning accelerators. It enables the accelerators to be developed by people with specific domain knowledge but no hardware or machine learning background. Through avoiding locks in data access, low-precision arithmetic, parallel programming language, and design space exploration, the project helps provide high performance, high productivity, and high efficient machine learning implementations.
Professor Yang Guangwen, an international authority in the field of supercomputing, introduced the system architecture and applications of the Sunway TaihuLight supercomputer. Professor Yang shared details of the deep learning platform and optimization progress of the Sunway computing system. When he described the long-term strategies of the Sunway TaihuLight supercomputer, the plan includes adding computational acceleration chip for AI applications. This coincides with the AI chip development direction in Corerain Technology. The two parties will further discuss the opportunities of cooperation.
In addition to the above academic key notes, Skysaga Capital Founding Partner Gary Yang Ge and Tencent Cloud Artificial Intelligence Director Wang Lei also shared the interests of venture capital and Tencent's roadmap in the field of artificial intelligence. In the roundtable forum on platform-level AI technology in AI ecosystem construction, these specific topics were further discussed.