信息存储和逻辑运算相融合的忆阻神经形态系统及其应用

With the dramatic increase in the amount of information, and higher information processing requirement, more intelligent and more diminutive information processing system are expected urgently. With regard to the fact that memristor can memory information as a human brain, it is expected to change the way of modern intelligent information processing completely. The project combines state-of-the-art circuit technology and biological science achievements to construct bionic intelligent information processing systems, in which information processing and storing are set in a single unit: (1) Designing binary-valued storage, multi-valued storage and analog storage of memristor-based memory cells using resistance random access memory (RRAM) technology; (2) Building memristor-based logical units to achieve Boolean operations and complex logic operations according to the switching characteristics of memristor; (3) importing biological intelligence learning rule to the integrate storage and logical function units organically to construct information processing storage units that are similar to biological brain neurons; (4) investigating the applications of the system in the logic operation, image processing, video tracking and so on. The system has the advantages of intelligence, miniaturization and low power consumption, so this project may effectively promote the integration of storage technology and information processing technology. The achievements of this project will provide important theoretical and practical foundations for developing intelligent information processing system. Studying in this field is still immature in the world, which has promising theoretical value and extremely competitive application prospects.

随着信息量的急剧增长和信息处理要求的不断提高，人们迫切需要更加智能化和微型化的信息处理系统。忆阻器具有类似于人类大脑的“记忆”功能，有望彻底改变现有的信息处理方式。本项目将电路和生命科学领域的最新研究成果相结合，研究信息存储和逻辑运算相融合的神经形态系统：利用阻变存储技术，构建二值、多值和模拟的忆阻存储单元；探究忆阻开关特性，构建逻辑运算单元，实现布尔逻辑和数值运算；引入生物智能学习规则，构造仿生神经细胞单元，实现存储和运算的有机融合；采用交叉阵列和三维堆叠结构，构建适合于超大规模集成电路实现的新型神经形态系统，并探讨其在逻辑运算、图像处理和视频跟踪等方面的应用。该系统具有智能化、微型化和低功耗等优势。研究成果将有效推进信息存储和处理技术的有机融合，为开发下一代智能信息处理系统提供重要的理论和实践依据。本项目拟开展的工作在国际上还处于萌芽状态，具有重要的理论研究价值和极具竞争力的应用前景。

根据项目任务书，本项目将电路和生命科学领域的最新研究成果相结合，研究了信息存储和逻辑运算相融合的神经形态系统。（1）利用阻变存储技术，构建二值、多值和模拟的忆阻存储单元；（2）探究忆阻开关特性，构建逻辑运算单元，实现布尔逻辑和数值运算；（3）引入生物智能学习规则，构造仿生神经细胞单元，实现存储和运算的有机融合；（4）采用交叉阵列和三维堆叠结构，构建适合于超大规模集成电路实现的新型神经形态系统，并探讨其在逻辑运算、图像处理和视频跟踪等方面的应用。该系统具有智能化、微型化和低功耗等优势。研究成果有效推进了信息存储和处理技术的优惠融合，为开发下一代智能信息处理系统提供重要的理论和实践依据。本项目开展的工作在国际上还处于萌芽状态，具有重要的理论研究价值和极具竞争力的应用前景。在项目研究期间圆满完成了预定的研究目标，我们在IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, IET CIRCUITS DEVICES & SYSTEMS, MATHEMATICAL PROBLEMS IN ENGINEERI NGIEEE和中国科学F辑: 信息科学等国内外知名杂志上发表学术论文65篇，其中SCI检索31篇，EI检索15篇，申请国家发明专利9项。2名博士研究生和16名硕士研究生已经在此项目中获益。

内容获取失败，请点击重试