An adaptive synaptic array using Fowler-Nordheim dynamic analog memory.
An adaptive synaptic array using Fowler-Nordheim dynamic analog memory.
复制标题
DOI：
10.1038/s41467-022-29320-6
复制
发表时间：
2022-03-29
影响因子：
16.6
通讯作者：
Chakrabartty S
中科院分区：
文献类型：
--
作者：
Mehta D;Rahman M;Aono K;Chakrabartty S
在本文中，我们提出了一种自适应突触阵列，可用于提高训练机器学习 (ML) 系统的能源效率。突触阵列由一组模拟存储元件组成，每个模拟存储元件本身就是一个微尺度动态系统，在其时间状态轨迹中存储信息。然后通过系统级学习算法对状态轨迹进行调制，以便将整体轨迹引导至最佳解决方案。我们证明，状态轨迹调制所需的外在能量可以与神经网络学习的动态相匹配，从而显着减少机器学习训练期间内存更新的能量耗散。因此，所提出的突触阵列可能对解决人工智能（AI）系统中观察到的训练和推理阶段之间的能源效率不平衡问题产生重大影响。虽然机器学习模型的应用已经取得了巨大进步，但在传统计算硬件上训练它们是一个能源密集型过程。在这里，Mehta 等人提出了一种自适应突触阵列，可显着提高训练的能量效率。
In this paper we present an adaptive synaptic array that can be used to improve the energy-efficiency of training machine learning (ML) systems. The synaptic array comprises of an ensemble of analog memory elements, each of which is a micro-scale dynamical system in its own right, storing information in its temporal state trajectory. The state trajectories are then modulated by a system level learning algorithm such that the ensemble trajectory is guided towards the optimal solution. We show that the extrinsic energy required for state trajectory modulation can be matched to the dynamics of neural network learning which leads to a significant reduction in energy-dissipated for memory updates during ML training. Thus, the proposed synapse array could have significant implications in addressing the energy-efficiency imbalance between the training and the inference phases observed in artificial intelligence (AI) systems. While great progress has been made in the applications of machine learning models, training them on conventional computing hardware is an energy intensive process. Here, Mehta et al present an adaptive synaptic array offering considerable improvements in the energy efficiency of training.
登录
查看更多内容
影响因子：
5.4
作者：
Chakrabartty, Shantanu;Cauwenberghs, Gert
通讯作者：
Cauwenberghs, Gert
影响因子：
3.4
作者：
Khvalkovskiy, A. V.;Apalkov, D.;Krounbi, M.
通讯作者：
Krounbi, M.
影响因子：
3.1
作者：
Gu, Xuefeng;Wan, Zhe;Iyer, Subramanian S.
通讯作者：
Iyer, Subramanian S.
影响因子：
16.6
作者：
Mehta D;Aono K;Chakrabartty S
通讯作者：
Chakrabartty S
DOI：
10.1109/tnnls.2017.2778940
发表时间：
2018-10-01
影响因子：
10.4
作者：
Merrikh-Bayat, Farnood;Guo, Xinjie;Strukov, Dmitri B.
通讯作者：
Strukov, Dmitri B.