摘要
针对人工智能的计量校准,阐述了其紧迫性、必要性以及可行性。提出了2种不同的人工智能计量解决方案。其一为基础方式,从智能的7个维度出发,分别进行语言智能、数理逻辑智能、空间智能、肢体运动智能、音乐智能、人际关系智能、内省智能的计量校准研究,然后,针对同时含有两种以上智能的多元智能系统,进行面向具体任务目标的综合加权,给出面对具体任务目标的智能水平定量评价结果。其二为工程方式,从每一个具体而明确的人工智能系统入手,依据愿景目标确立评价指标体系,开展计量校准研究,最终以定量量化方式,评价系统的分项与综合智能水平。其主导思想是以定量量化方式,评价任意智能系统的智能水平。
For the measurement and calibration of artificial intelligence,to describe its urgency,necessity and feasibility. Two different artificial intelligence metering solutions are proposed. The first method is based on the seven dimensions of intelligence,and performs linguistic intelligence,mathematical logic intelligence,spatial intelligence,physical intelligence,music intelligence,interpersonal relationship intelligence,and introspective intelligence. Two or more intelligent multiple intelligence systems perform comprehensive weighting for specific task objectives,and give a quantitative evaluation result of the intelligence level to specific task objectives. The second is the engineering method,starting with each specific and clear artificial intelligence system,establishing an evaluation system according to the ideal goal,conducting a measurement calibration study,and finally evaluating the system ’s sub-items and comprehensive intelligence level in a quantitative manner. Its main idea is to evaluate the intelligence level of any intelligent system in a quantitative way.
作者
梁志国
姜延欢
LIANG Zhi-guo;JIANG Yan-huan(National Key Laboratory of Science and Technology on Metrology&Calibration,Changcheng Institute of Metrology and Measurement,Beijing 100095,China)
出处
《计量学报》
CSCD
北大核心
2021年第1期78-84,共7页
Acta Metrologica Sinica
关键词
计量学
人工智能
校准
评价
metrology
artificial intelligence
calibration
evaluation
