IEEE C/DC 2941 Working Group
IEEE C/DC AI Model Representation, Compression, Distribution and Management Working Group

IEEE AI Model Representation, Compression, Distribution and Management Working Group Officially Established

IEEE数据压缩标准委员会AI模型国际标准工作组正式成立

IEEE AI Model Representation, Compression, Distribution and Management Working Group Officially Established

以鹏城实验室与北京大学、华为技术有限公司、北京百度网讯科技有限公司等国内一流院校、科研单位、科技企业的科研人员为主要初始成员的AI模型标准工作组,于近日通过全球最大非营利专业技术学会–电气和电子工程师协会新标准评审委员会(IEEE NesCom)的审议,正式成为IEEE数据压缩国际标准委员会(IEEE Data Compression Standard Committee, 简称 IEEE C/DC)下的国际标准工作组之一。

The AI Model Representation, Compression, Distribution and Management (C/DC/AIM) Working Group, with its main founding members from PCL, Peking University, Baidu and other top universities, scientific research institutions, and technology enterprises in China, has recently passed the review of the IEEE New Standards Committee (IEEE NesCom), and officially became one of the working groups under the IEEE Data Compression Standard Committee (IEEE C/DC).

AI模型标准工作组全称为AI模型的表示、压缩、分配和管理工作组(AI Model Representation, Compression, Distribution and Management,简称C/DC/AIM),计划推出一系列标准来定义AI开发接口、AI模型可互操作表示、编码格式和模型封装格式,以用于高效的AI模型推理、存储、分发和管理。工作组将结合《信息技术 神经网络表示与模型压缩》系列国家标准中的部分工作,开展一系列深入研究,为AI模型的推理、压缩与解压缩、编码和打包提供有效的工具集,打破AI模型在不同计算架构和算法平台之间的壁垒,同时达到降低存储成本、保护敏感信息的目标。该系列标准可广泛运用于端计算和云计算场景,可用于神经网络模型等的研制、开发、测试评估和应用,可服务包括AI教育、AI医疗、智慧城市在内的人工智能热门应用领域。

The IEEE C/DC/AIM plans to introduce a series of standards to define AI development interfaces, AI model interoperable representations, coding formats, and model encapsulated formats for efficient AI model inference, storage, distribution, and management. The IEEE C/DC/AIM will combine part of the work in the series of China’s national standards of “Artificial intelligence—Neural Network Representation and Model Compression” to carry out in-depth studies and provide an effective toolset for the inference, compression, decompression, coding and packaging of AI models. This work aims at breaking the barriers of AI models between different computing architectures and algorithmic frameworks, while achieving the goal of reducing storage costs and protecting business-sensitive information. This series of standards not only can be widely used in client computing and cloud computing scenarios, but also can be adopted for research, development, testing, evaluation, and application of neural network models in a wide range of AI applications including AI education, AI healthcare, and smart cities.

AI模型国际标准工作组采用个人成员方式,面向全球研究人员开放。工作组希望汇聚全球人工智能及信息技术领域专家一同开展研究,广泛征集和了解行业需求,从而开展真正有效且有意义的国际标准研究工作。

The IEEE C/DC/AIM adopts individual membership and is open to global researchers, with an objective to bring together experts in the field of artificial intelligence and information technology from across the world to jointly conduct research, extensively collect and understand the needs of the industry, and carry out more effective and useful research on international standards.