Neural Networks, Hypersurfaces, and the Generalized Radon Transform.
Neural Networks, Hypersurfaces, and the Generalized Radon Transform.
复制标题
神经网络、超曲面和广义氡变换。
DOI：
10.1109/msp.2020.2978822
复制
发表时间：
2020
影响因子：
14.9
通讯作者：
Rohde,GustavoK
中科院分区：
文献类型：
--
作者：
Kolouri,Soheil;Yin,Xuwang;Rohde,GustavoK
人工神经网络（ANN）长期以来一直被用作数学建模方法，最近在科学和技术中发现了许多应用，包括计算机视觉，信号处理和机器学习[1]，仅举几例。虽然存在显着的函数近似结果[2]，但理论解释尚未赶上新的发展，特别是关于（深度）分层学习。因此，许多疑问往往伴随着NN从业者，如一个应该使用多少层？不同激活函数的作用是什么？池化有什么影响？和其他许多人.
Artificial neural networks (ANNs) have long been used as a mathematical modeling method and have recently found numerous applications in science and technology, including computer vision, signal processing, and machine learning [1], to name a few. Although notable function approximation results exist [2], theoretical explanations have yet to catch up with newer developments, particularly with regard to (deep) hierarchical learning. As a consequence, numerous doubts often accompany NN practitioners, such as How many layers should one use? What is the effect of different activation functions? What are the effects of pooling? and many others.