Adaptive Deconvolutional Networks for Mid and High Level Feature Learning(阅读)

作者在原代价函数做了进一步的改进,相比上一篇DN的不同就是直接对原图进行代价函数的重构,并不是对上一层输出重构。这篇文章可以了解学习一下。

Adaptive Deconvolutional Networks for Mid and High Level Feature Learning(阅读)_第1张图片Adaptive Deconvolutional Networks for Mid and High Level Feature Learning(阅读)_第2张图片Adaptive Deconvolutional Networks for Mid and High Level Feature Learning(阅读)_第3张图片

你可能感兴趣的:(Adaptive Deconvolutional Networks for Mid and High Level Feature Learning(阅读))