arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.05262math.NAcs.NA
收藏

用于函数逼近与神经网络训练的Fredholm积分方程

Fredholm integral equations for function approximation and the training of neural networks

Patrick Gelß, Aizhan Issagali, Ralf Kornhuber

AI总结:

本研究提出一种基于第一类Fredholm积分方程近似最小二乘解的函数逼近与神经网络训练新方法,结合Ritz-Galerkin离散化等技术,经监督学习实验验证其性能可与前沿神经网络方法媲美。

摘要:

我们提出了一种新颖且数学上透明的函数逼近与大规模高维神经网络训练方法,该方法基于通过Ritz-Galerkin离散化、Tikhonov正则化和tensor-train方法求解相关第一类Fredholm积分方程的近似最小二乘解。在回归和分类类型的监督学习问题上的实际应用证实,所得算法与基于神经网络的最新方法具有竞争力。

英文摘要:

We present a novel and mathematically transparent approach to function approximation and the training of large, high-dimensional neural networks, based on the approximate least-squares solution of associated Fredholm integral equations of the first kind by Ritz-Galerkin discretization, Tikhonov regularization and tensor-train methods. Practical application to supervised learning problems of regression and classification type confirm that the resulting algorithms are competitive with state-of-the-art neural network-based methods.

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。

↑