Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

算法重新训练不收敛 #8

Open
aliendaniel opened this issue Dec 10, 2019 · 2 comments
Open

算法重新训练不收敛 #8

aliendaniel opened this issue Dec 10, 2019 · 2 comments

Comments

@aliendaniel
Copy link

@taosir 您好,我用分别用tf2 keras、paddlepaddle、pytorch重新复现了您的算法,用hwdb1.1数据训练模型,图像处理的过程为取灰度,resize到(64,64,1),标准化到01区间,发现均不收敛loss始终在8.2左右,可否告知下训练前是否对图像做了其它的调整。

@taosir
Copy link
Owner

taosir commented Dec 11, 2019

没有做特别的处理,只是一些简单的图片旋转,缩放,亮度变化。
不收敛的话考虑一下:
1、数据做归一化
2、忘了做数据预处理。忘了使用正则化
3、Batch Size设的太大。学习率设的不对
4、最后一层的激活函数用的不对
5、梯度爆炸
6、参数初始化调整
7、网络太深
8、隐藏层神经元数量调整

@gy-hhhh
Copy link

gy-hhhh commented Mar 13, 2020

@aliendaniel 您好,请问您目前问题解决了吗?我也出现了同样的问题,loss在8.2左右不下降,如果您已经解决,麻烦您说一下问题出在哪儿,谢谢。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants