在机器学习里,经常会看到两个词,data fidelity term, regularization(prior) term.
例如,在image restoration中,我们需要最小化如下的object function(MAP, 最大后验概率),则:
data fidelity term:
regularization(prior) term:
那么两项的作用是什么呢?
The fidelity term guarantees the solution accords with the degradation process,
翻译:数据保真项保证结果符合降质过程
while the regularization term enforces desired property of the output.
翻译:正则(先验)项对输出进行增强
Since IR is an ill-posed inverse problem, the prior which is also called regularization needs to be adopted to constraint the solution space
翻译:因为image restoration是病态解问题,所以用正则(先验)项限制解空间