摘要
仿人智能的核心在于对人脑的深入研究与不断模仿。对仿人智能领域的研究进行了综合评述,并对进一步发展方向进行了分析。首先概述了大脑结构及处理机制研究,大规模突触级别脑结构图问世,推动对脑结构与功能的深入刻画;然后讨论了最新的类脑计算模型和算法研究成果,类脑计算成为通用人工智能的基石,具有极其广阔的应用前景;最后介绍了计算与存储一体的计算架构研究进展,有着低功耗和快速计算特点的神经拟态芯片逐渐成为未来超大规模人工智能应用的基石。基于上述分析,阐释仿人智能未来发展趋势。综述表明,通过模仿生物神经网络实现机器智能已经成为一条十分重要的研究路线,未来它甚至有可能突破人工智能和生物智能的天花板。
The core of simulation-human intelligence lies in the deep research and imitation of human brain.This paper makes a comprehensive review of the research in the field of simulation-human intelligence,and analyzes the further development directions.This paper firstly summarizes the mechanism of brain structure and processing mechanism,and meanwhile the large-scale synaptic-level brain structure graph is published,which promotes the in-depth description of brain’s structure and function.Then this paper discusses the latest research achievements of brain-like computing model and algorithm.Nowadays,brain-like computing has become the cornerstone of general artificial intelligence(GAI)and has extremely broad application prospects.Finally,this paper introduces the research progress of computing architecture integrating computing and storage.In recent years,the neural-mimicry chip with the abilities of low power consumption and fast computing has gradually become the cornerstone of super large-scale AI applications in the future.Based on the aforementioned analysis,the future development trend of simulation-human intelligence is introduced.The proposed survey shows that,it has become a very important research route to realize machine intelligence by imitating biological neural network,and it may even break through the ceiling of artificial intelligence(AI)and biological intelligence(BI)in the future.
作者
李芳
祝翠琴
LI Fang;ZHU Cuiqin(Institute of Scientific and Technical Information of China,Beijing 100038,China;School of Computer,Beijing Institute of Technology,Beijing 100081,China)
出处
《无人系统技术》
2021年第2期8-13,共6页
Unmanned Systems Technology
关键词
仿人智能
人工智能
脑结构图
类脑计算
神经拟态芯片
仿生学
Simulation-Human Intelligent
Artificial Intelligence
Brain Structure Graph
Brain-Like Computing
Neural-Mimicry Chip
Bionics
