线性对称自选14位运动分子记忆器
Deepak Sharma1, Santi Prasad Rath1, Bidyabhusan Kundu1
1Centre for Nano Science and Engineering, Indian Institute of Science, Bangalore, India.
Nature
|September 11, 2024
概括
研究人员开发了一种用于人工智能 (AI) 的新型模拟分子记忆器. 这种高分辨率的设备可以有效地训练神经网络和信号处理,显著提高神经形态计算能力.
科学领域:
- 材料科学
- 计算机工程
- 纳米技术
背景情况:
- 人工智能依赖于资源密集型数据中心,限制了可访问性.
- 目前的神经形态硬件提供了能源效率, 但缺乏像训练这样复杂的人工智能任务的准确性.
- 高分辨率计算对于信号处理,神经网络训练和自然语言处理至关重要.
研究的目的:
- 为先进的人工智能应用引入一种具有高分辨率的新型模拟分子记忆器.
- 展示神经形态平台的简化和高效的重量更新机制.
- 开发一个没有选择器的横杆引擎,用于快速的向量矩阵乘法.
主要方法:
- 开发具有14位分辨率的Ru复杂模拟分子记忆器.
- 使用精确的动力控制来实现稳定的分子电子状态和不同的模拟导电水平.
- 构建一个没有选择器的64x64横杆基点产品引擎.
主要成果:
- 通过线性和对称更新实现了16,520个不同的模拟导电水平.
- 在单个时间步骤中启用矢量矩阵乘法,包括里埃变换.
- 显示信号噪声比超过73dB,比数字计算机低460倍的能耗.
结论:
- 分子记忆器在AI硬件的分辨率和效率上提供了显著的进步.
- 这项技术可以扩展神经形态计算的应用范围.
- 开发的横杆加速器有潜力扩大从云端到边缘的数字电子.
相关概念视频
MOS Capacitor
732
A Metal-Oxide-Semiconductor (MOS) capacitor is a fundamental structure used extensively in semiconductor device technology, particularly in the fabrication of integrated circuits and MOSFETs (metal-oxide-semiconductor field-effect transistors). The MOS capacitor consists of three layers: a metal gate, a dielectric oxide, and a semiconductor substrate.
The metal gate is typically made from highly conductive materials such as aluminum or polysilicon. Beneath the metal gate lies a thin layer of...
The metal gate is typically made from highly conductive materials such as aluminum or polysilicon. Beneath the metal gate lies a thin layer of...
732
Potentiometry: Membrane Electrodes
498
Membrane electrodes, also known as p-ion electrodes, use membranes that selectively interact with free analyte ions, generating a potential difference across the membrane. The resulting membrane potential, known as the asymmetry potential, is not zero even when analyte concentrations on both sides of the membrane are equal. The membrane's response is typically not selective to a single analyte but proportional to the concentration of all ions in the sample solution capable of interacting at...
498
The Resting Membrane Potential
131.5K
Overview
131.5K


