【图像质量评价】No-reference Image Quality Assessment with Deep Convolutional Neural Networks - 2016 ---3区

一、直接引用

1、P2：“[15] demonstrated that in a deep CNN that ReLUs enable the network to train

several times faster compared to using tanh units.”

2、P2: “The convolution filter in CNN is a Generalized Linear Model (GLM) for the underlying data patch. In [16], Lin et al. Argue that the level of abstraction is low with GLM, which means the learned features by traditional CNN filters is variant to the variants of the same concept.”

P2: “Lin et al. propose to replace the GLM with ab“micro network” structure which is a general nonlinear function approximator and instantiate it using MLP.”

4、P3: “Similar with [11], we assign each