arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.12975cs.CVcs.LGstat.ML
收藏

用梯度下降训练形态学神经网络:若干理论见解

Training morphological neural networks with gradient descent: some theoretical insights

Mines Paris(巴黎矿业学院);PSL University(巴黎文理研究大学)

Samy Blusseau

AI总结:

本文从Bouligand导数视角分析梯度下降训练多层形态学神经网络的困难,并给出初始化与学习率的初步理论指导。

摘要:

形态学神经网络或形态学层可以成为推动数学形态学进展的有力工具,无论是在完备格算子表示等理论方面,还是在图像处理流程的开发中。然而,当这些架构包含超过少数几个形态学层时,至少在基于梯度下降优化算法的流行机器学习框架中,它们往往难以训练。本文从非光滑优化中Bouligand导数的概念出发,研究基于微分的方法和反向传播应用于形态学网络的潜力与局限。我们提供了若干见解和初步的理论指导,特别是关于初始化和学习率方面的建议。

英文摘要:

Morphological neural networks, or layers, can be a powerful tool to boost the progress in mathematical morphology, either on theoretical aspects such as the representation of complete lattice operators, or in the development of image processing pipelines. However, these architectures turn out to be difficult to train when they count more than a few morphological layers, at least within popular machine learning frameworks which use gradient descent based optimization algorithms. In this paper we investigate the potential and limitations of differentiation based approaches and back-propagation applied to morphological networks, in light of the non-smooth optimization concept of Bouligand derivative. We provide insights and first theoretical guidelines, in particular regarding initialization and learning rates.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑