Tensorboard的零值分数应如何解释?

ree*_*106 4 tensorflow tensorboard google-cloud-ml-engine

I am running a cloud ML engine job and my tensorboard plots are showing the fraction of zero values for my hidden layers steadily increasing towards 1 as the number of steps increases. How should this plot be interpreted? I believe it is a good thing as more zero values would suggest that the model is getting more "certain" about the predictions that it is making.

张量板:零值的隐藏层分数

rha*_*l80 6

通常,这意味着您的正则化技术和/或激活功能正在将激活强制为零。您尚未共享模型的详细信息,但这在使用辍学(尤其是relu激活功能)时很常见。

具有大量零激活的模型倾向于更好地泛化,因此具有更好的准确性。

如果您需要更多详细信息,请参阅有关dropout的JMLR论文

我确实要注意,激活至少为零有时是不好的,至少对于ReLU激活功能而言。基本上,他们可以毫不客气地“死”。因此,如果看到不良的模型质量,请当心。更多信息在这里