Brief introduction to the Model and Algorithm Working Group:
The Model and Algorithm Working Group was established on 6 August 2020, during the first planetary meeting of the National Information Security Standardization Technical Committee’s Subcommittee for Artificial Intelligence (TC 28/SC 42). Its work focuses on the review and analysis of AI fundamental models and algorithms, specifications for generic models and algorithms in key areas, research on AI development framework and open platforms, as well as related standardisation to support industrial application.
Group leader: Wu Wenjun from Beihang University
Deputy group leader: Ma Yanjun from Beijing Baidu Netcom Science and Technology Co.,Ltd.
Deputy group leader: Mei Jingqing from Beijing Megvii Technology Limited
(Secretariat Contact: Ma Chenghao 16600049001 / 010-64102859)
Achievements in AI models and algorithms:
- T/CESA 1026—2018 Artificial intelligence – Assessment specification for deep learning algorithms
- T/CESA 1034—2019 Information technology – Artificial intelligence – Sample size and algorithm requirements for few-shot learning
- T/CESA 1036—2019 Information technology – Artificial intelligence – Quality elements and testing methods of machine learning model and system
- T/CESA 1037—2019 Information technology – Artificial intelligence – Framework and functional requirements of system for machine learning
- T/CESA 1040—2019 Information technology – Artificial intelligence – Code of practice for data annotation of machine learning
The following table summarises the ongoing work of the Working Group:
| Category | Name |
| Research report | Report on the fairness and supervision of artificial intelligence algorithms |
| Research report | White paper on technical development and industrial application of artificial intelligence models and algorithms |
| Standard | IEEE P3142 Recommended practice on distributed training and inference for large-scale deep learning models |
| Standard | Information technology – Artificial intelligence – Technical requirements and evaluation indicators for adaption of multi-platform deep learning frameworks |
| Standard | Information technology – Artificial intelligence – Guidelines for model management |
| Standard | Artificial intelligence – Technical requirements for Application Programming Interface (API) of deep learning inference engine |
| Standard | Artificial intelligence – Functional and technical requirements for deep learning framework |
| Standard | Series standards of information technology – Neural network representation and model compression |
| Standard | Artificial intelligence – Code of practice for data annotation of machine learning |
| Standard | Artificial intelligence – Specification for machine learning systems |
| Standard | Artificial intelligence – Specification for the assessment of deep learning algorithms |
| Standard | Artificial intelligence – Technical requirements for algorithm management of multi-algorithm application system |
