最大池层与具有步幅性能的卷积
Max pool layer vs Convolution with stride performance
在大多数架构中,conv 层之后是池化层(max / avg 等)。由于那些池化层只是选择前一层的输出(即 conv),我们是否可以只使用步幅为 2 的卷积并期望获得相似的精度结果并减少过程需求?
是的,可以做到。它在论文 'Striving for simplicity: The all convolutional net'
https://arxiv.org/pdf/1412.6806.pdf 中有解释。引自论文:
'We find that max-pooling can simply be replaced by a convolutional
layer with increased stride without loss in accuracy on several image
recognition benchmarks'
在大多数架构中,conv 层之后是池化层(max / avg 等)。由于那些池化层只是选择前一层的输出(即 conv),我们是否可以只使用步幅为 2 的卷积并期望获得相似的精度结果并减少过程需求?
是的,可以做到。它在论文 'Striving for simplicity: The all convolutional net'
https://arxiv.org/pdf/1412.6806.pdf 中有解释。引自论文:
'We find that max-pooling can simply be replaced by a convolutional layer with increased stride without loss in accuracy on several image recognition benchmarks'