A guide to convolution arithmetic for deep learning
A guide to convolution arithmetic for deep learning
复制标题
DOI：
--
复制
发表时间：
2016-03
期刊：
影响因子：
--
通讯作者：
Vincent Dumoulin;Francesco Visin
中科院分区：
文献类型：
--
作者：
Vincent Dumoulin;Francesco Visin
我们介绍了一个指南，以帮助深度学习从业者理解和操作卷积神经网络架构。该指南阐明了卷积层、池化卷积层和转置卷积层的各种属性（输入形状、内核形状、零填充、步幅和输出形状）之间的关系，以及卷积层和转置卷积层之间的关系。各种情况下的关系，并说明，以使他们直观。
We introduce a guide to help deep learning practitioners understand and manipulate convolutional neural network architectures. The guide clarifies the relationship between various properties (input shape, kernel shape, zero padding, strides and output shape) of convolutional, pooling and transposed convolutional layers, as well as the relationship between convolutional and transposed convolutional layers. Relationships are derived for various cases, and are illustrated in order to make them intuitive.