在边缘支持AI:一个强大的,基于memristor的二元化神经网络,具有近内存计算和微型太阳能电池
Fadi Jebali1, Atreya Majumdar2, Clément Turck2
1Aix-Marseille Université, CNRS, Institut Matériaux Microélectronique Nanosciences de Provence, Marseille, France.
Nature communications
|January 25, 2024
概括
这项研究介绍了一个强大的二元化神经网络,使用memristors和太阳能发电来实现节能的人工智能 (AI). 它甚至在低光条件下也展示了功能性AI操作,为自动供电的智能传感器铺平了道路.
科学领域:
- 材料科学 材料科学 材料科学
- 计算机工程 计算机工程
- 人工智能的人工智能
背景情况:
- 基于memristor的神经网络提供了节能的人工智能 (AI) 和自动供电运行的潜力.
- 在memristor网络中的模拟内存计算需要稳定的电力,与不可靠的能量收获器相冲突.
研究的目的:
- 开发一个强大的,节能的基于memristor的神经网络,用于由能源收获器驱动的AI应用程序.
- 通过采用数字近内存计算方法,克服使用不稳定的电源的模拟计算的局限性.
主要方法:
- 一个二元化神经网络的制造,拥有32,768个memristor.
- 与微型宽带间隙太阳能电池集成供电.
- 实现一个数字近内存计算架构与补充编程和逻辑感应放大器.
主要成果:
- 该电路在高光照明下表现出与实验室长板电源相匹配的推断性能.
- 功能操作在低光照明下保持,过渡到近似计算,精度略有降低.
- 模拟表明,低光下错误分类主要涉及难以分类的图像.
结论:
- 开发的系统为自动供电的人工智能系统奠定了基础.
- 这项技术可以创建用于健康,安全和环境监测的智能传感器.
相关概念视频
Non-equilibrium in the Cell
4.4K
An important concept in studying metabolism and energy is that of chemical equilibrium. Most chemical reactions are reversible. They can proceed in both directions, releasing energy into their environment in one direction, and absorbing it from the environment in the other direction. The same is true for the chemical reactions involved in cell metabolism, such as the breaking down and building up of proteins into and from individual amino acids, respectively. Reactants within a closed system...
4.4K
P-N junction
534
A p-n junction is formed when p-type and n-type semiconductor materials are joined together. At the interface of the p-n junction, holes from the p-side and electrons from the n-side begin to diffuse into the opposite sides due to the concentration gradient. This diffusion of carriers leads to a region around the junction where there are no free charge carriers, known as the depletion region. The charge density within the depletion region for the n-side and p-side can be described by the...
534


