神经网络高阶技巧4--关于SpatialDropout1D

最新推荐文章于 2024-05-30 15:27:22 发布

ssswill

最新推荐文章于 2024-05-30 15:27:22 发布

阅读量3.4k

点赞数 1

分类专栏： DL

本文链接：https://blog.csdn.net/ssswill/article/details/90054469

版权

参见：
https://blog.csdn.net/weixin_43896398/article/details/84762943

但是这里有一个疑问。文中提到了noise_shape，但是在官方文档，以及keras.layers.SpatialDropout1D()的函数说明里貌似也没有，或者说它的默认值里哪个轴为1呢？
我们在jupyter lab中输入

??SpatialDropout1D

Init signature: SpatialDropout1D(rate, **kwargs)
Source:        
class SpatialDropout1D(Dropout):
    """Spatial 1D version of Dropout.

    This version performs the same function as Dropout, however it drops
    entire 1D feature maps instead of individual elements. If adjacent frames
    within feature maps are strongly correlated (as is normally the case in
    early convolution layers) then