学习笔记GAN002:DCGAN

Ian J. Goodfellow 论文：https://arxiv.org/abs/1406.2661

两个网络：G(Generator)，生成网络，接收随机噪声Z，通过噪声生成样本，G(z)。D(Dicriminator)，判别网络，判别样本是否真实，输入样本x，输出D(x)代表x真实概率，如果1,100%真实样本，如果0,代表不可能是真实样本。

训练过程，生成网络G尽量生成真实样本欺骗判别网络D，判别网络D尽量把G生成样本和真实样本分别开。理想状态下，G生成样本G(z)，使D难以判断真假，D(G(z))=0.5。此时，生成模型G，可以用来生成样本。

数学公式：minG maxDV(D,G)=Ex~pdata(x)[logD(x)]+Ez~pz(z)[log(1-D(G(z)))]
二项式。x真实样本，z输入G网噪声，G(z) G网生成样本。D(x) D网判断真实样本是否真实概率，越接近1越好。D(G(z)) D网判断G网生成样本真实概率。G网，D(G(z))尽可能大，V(D,G)变小，min_G。D网，D(x)越大，D(G(x))越小，V(D,G)越大，max_D。
1､x sampled from data -> Differentiable function D -> D(x) tries to be near 1
2､Input noise z -> Differntiable function G -> x sampled from model -> D -> D tries to make D(G(z)) near 0,G tries to make D(G(z)) near 1

随机梯度下降法训练D、G。

Algorithm 1 Minibatch stochastic gradient descent training of genegative adversarial nets. The number of steps to apply to the discriminator,k,is a hyperparameter, we used k = 1,the least expensive option, in our experiments.
for number of training iterations do
    for k steps do
        Sample minibatch of m noise samples {z(1),...,z(m)} from noise prior pg(z)
        Sample minibatch of m examples {x(1),...,x(m)} from data generating distribution pdata(x)
        Update the discriminator by ascending its stochastic gradient:
    end for
    Sample mninbatch of m noise samples {z(1),...,z(m)} from noise prior pg(z)
    Update the generator by descending its stochastic gradient:
end for
The gradient-based updates can use any standard gradient-based learning rule.We used momenttum in our experiments.

第一步训练D,V(G,D)越大越好，上升(增加)梯度(ascending)。第二步训练G，V(G,D)越小越好，下降(减少)梯度(descending)。交替进行。

DCGAN原理。https://arxiv.org/abs/1511.06434 。Alec Rad