当前位置:科学网首页 > 小柯机器人 >详情
科学家提出深度物理神经网络的无反向传播训练方法
作者:小柯机器人 发布时间:2023/11/24 23:44:08

近日,瑞士洛桑联邦理工学院的Romain Fleury及其研究小组取得一项新进展。经过不懈努力,他们提出深度物理神经网络的无反向传播训练方法。相关研究成果已于2023年11月23日在《科学》期刊上发表。

该研究团队提出了一种简化的深度神经网络架构,其通过物理局部学习(PhyLL)算法增强,实现深度物理神经网络的监督和无监督训练,而无需深入理解非线性物理层的性质。在元音和图像分类实验中,研究人员训练了多种基于波的物理神经网络,体现了该方法的通用性。此方法在提升训练速度、增强鲁棒性的同时,还通过消除对系统建模的需求来减少数字计算,进而降低功耗,展现出相较于其他硬件感知训练方案的优势。

据悉,近年来深度学习在视觉和自然语言处理领域的成功主要归因于更大的模型。然而,这种成功带来的问题是能耗和可扩展性的挑战。目前,数字深度学习模型的训练主要依赖于反向传播算法,但这种算法并不适合物理实现。

附:英文原文

Title: Backpropagation-free training of deep physical neural networks

Author: Ali Momeni, Babak Rahmani, Matthieu Malléjac, Philipp del Hougne, Romain Fleury

Issue&Volume: 2023-11-23

Abstract: Recent successes in deep learning for vision and natural language processing are attributed to larger models but come with energy consumption and scalability issues. Current training of digital deep learning models primarily relies on backpropagation that is unsuitable for physical implementation. Here, we proposed a simple deep neural network architecture augmented by a physical local learning (PhyLL) algorithm, enabling supervised and unsupervised training of deep physical neural networks, without detailed knowledge of the nonlinear physical layer’s properties. We trained diverse wave-based physical neural networks in vowel and image classification experiments, showcasing our approach’s universality. Our method shows advantages over other hardware-aware training schemes by improving training speed, enhancing robustness, and reducing power consumption by eliminating the need for system modelling and thus decreasing digital computation.

DOI: adi8474

Source: https://www.science.org/doi/10.1126/science.adi8474

期刊信息
Science:《科学》,创刊于1880年。隶属于美国科学促进会,最新IF:63.714