Ho J., Jain A. and Abbeel P. Denoising diffusion probabilistic models. In Advances in Neural Information Processing Systems (NIPS), 2020.

[Page E. Approximating to the cumulative normal function and its inverse for use on a pocket calculator. Applied Statistics, vol. 26, pp. 75-76, 1977.]

Yerukala R., Boiroju N. K. Approximations to standard normal distribution function. Journal of Scientific and Engineering Research, vol. 6, pp. 515-518, 2015.

diffusion model和变分界的结合.

对抗鲁棒性上已经有多篇论文用DDPM生成的数据用于训练了, 可见其强大.

主要内容

Diffusion models

reverse process

从\(p(x_T) = \mathcal{N}(x_T; 0, I)\)出发:

\[p_{\theta}(x_{0:T}) := p(X_T) \prod_{t=1}^T p_{\theta}(x_{t-1}|x_t), \quad p_{\theta}(x_{t-1}|x_t) := \mathcal{N}(x_{t-1}; \mu_{\theta}(x_{t}, t), \Sigma_{\theta}(x_t, t)),
\]

注意这个过程我们拟合均值\(\mu_{\theta}\)和协方差矩阵\(\Sigma_{\theta}\).

这部分的过程逐步将噪声'恢复'为图片(信号)\(x_0\).

forward process

\[q(x_{1:T}|x_0) := \prod_{t=1}^{T}q(x_t|x_{t-1}), \quad q(x_t|x_{t-1}):= \mathcal{N}(x_t; \sqrt{1 - \beta_t} x_{t-1}, \beta_t I).
\]

其中\(\beta_t\)是可训练的参数或者人为给定的超参数.

这部分为将图片(信号)逐步添加噪声的过程.

变分界

对于参数\(\theta\), 很自然地我们希望通过最小化其负对数似然来优化:

\[\begin{array}{ll}
\mathbb{E}_{p_{data}(x_0)} \bigg[-\log p_{\theta}(x_0) \bigg]
&=\mathbb{E}_{p_{data}(x_0)} \bigg[-\log \int p_{\theta}(x_{0:T}) \mathrm{d}x_{0:T} \bigg] \\
&=\mathbb{E}_{p_{data}(x_0)} \bigg[-\log \int q(x_{1:T}|x_0)\frac{p_{\theta}(x_{0:T})}{q(x_{1:T}|x_0)} \mathrm{d}x_{0:T} \bigg] \\
&=\mathbb{E}_{p_{data}(x_0)} \bigg[-\log \mathbb{E}_{q(x_{1:T}|x_0)} \frac{p_{\theta}(x_{0:T})}{q(x_{1:T}|x_0)} \bigg] \\
&\le -\mathbb{E}_{p_{data}(x_0)}\mathbb{E}_{q(x_{1:T}|x_0)} \bigg[\log \frac{p_{\theta}(x_{0:T})}{q(x_{1:T}|x_0)} \bigg] \\
&= -\mathbb{E}_q \bigg[\log \frac{p_{\theta}(x_{0:T})}{q(x_{1:T}|x_0)} \bigg] \\
&= -\mathbb{E}_q \bigg[\log p(x_T) + \sum_{t=1}^T \log \frac{p_{\theta}(x_{t-1}|x_t)}{q(x_t|x_{t-1})} \bigg] \\
&= -\mathbb{E}_q \bigg[\log p(x_T) + \sum_{t=2}^T \log \frac{p_{\theta}(x_{t-1}|x_t)}{q(x_t|x_{t-1})} + \log \frac{p_{\theta}(x_0|x_1)}{q(x_1|x_0)} \bigg] \\
&= -\mathbb{E}_q \bigg[\log p(x_T) + \sum_{t=2}^T \log \frac{p_{\theta}(x_{t-1}|x_t)}{q(x_{t-1}|x_t, x_0)} \cdot \frac{q(x_{t-1}|x_0)}{q(x_t|x_0)} + \log \frac{p_{\theta}(x_0|x_1)}{q(x_1|x_0)} \bigg] \\
&= -\mathbb{E}_q \bigg[\log \frac{p(x_T)}{q(x_T|x_0)} + \sum_{t=2}^T \log \frac{p_{\theta}(x_{t-1}|x_t)}{q(x_{t-1}|x_t, x_0)} + \log p_{\theta}(x_0|x_1) \bigg] \\
\end{array}
\]

注: \(q=q(x_{1:T}|x_0)p_{data}(x_0)\), 下面另\(q(x_0) := p_{data}(x_0)\).

\[\begin{array}{ll}
\mathbb{E}_q [\log \frac{q(x_T|x_0)}{p(x_T)}]
&= \int q(x_0, x_T) \log \frac{q(x_T|x_0)}{p(x_T)} \mathrm{d}x_0 \mathrm{d}x_T \\
&= \int q(x_0) q(x_T|x_0) \log \frac{q(x_T|x_0)}{p(x_T)} \mathrm{d}x_0 \mathrm{d}x_T \\
&= \int q(x_0) \mathrm{D_{KL}}(q(x_T|x_0) \| p(x_T)) \mathrm{d}x_0 \\
&= \int q(x_{0:T}) \mathrm{D_{KL}}(q(x'_T|x_0) \| p(x'_T)) \mathrm{d}x_{0:T} \\
&= \mathbb{E}_q \bigg[\mathrm{D_{KL}}(q(x'_T|x_0) \| p(x'_T)) \bigg].
\end{array}
\]

\[\begin{array}{ll}
\mathbb{E}_q [\log \frac{q(x_{t-1}|x_t, x_0)}{p_{\theta}(x_{t-1}|x_t)}]
&=\int q(x_0, x_{t-1}, x_t) \log \frac{q(x_{t-1}|x_t, x_0)}{p_{\theta}(x_{t-1}|x_t)} \mathrm{d}x_0 \mathrm{d}x_{t-1}\mathrm{d}x_t\\
&=\int q(x_0, x_t) \mathrm{D_{KL}}(q(x_{t-1}|x_t, x_0)\| p_{\theta}(x_{t-1}|x_t)) \mathrm{d}x_0 \mathrm{d}x_t\\
&=\mathbb{E}_q\bigg[\mathrm{D_{KL}}(q(x'_{t-1}|x_t, x_0)\| p_{\theta}(x'_{t-1}|x_t)) \bigg].
\end{array}
\]

故最后:

\[\mathcal{L} := \mathbb{E}_q \bigg[
\underbrace{\mathrm{D_{KL}}(q(x'_T|x_0) \| p(x'_T))}_{L_T} +
\sum_{t=2}^T \underbrace{\mathrm{D_{KL}}(q(x'_{t-1}|x_t, x_0)\| p_{\theta}(x'_{t-1}|x_t))}_{L_{t-1}}
\underbrace{-\log p_{\theta}(x_0|x_1)}_{L_0}.
\bigg]
\]

损失求解

因为无论forward, 还是 reverse process都是基于高斯分布的, 我们可以显示求解上面的各项损失:

首先, 对于forward process中的\(x_t\):

\[\begin{array}{ll}
x_t
&= \sqrt{1 - \beta_t} x_{t-1} + \sqrt{\beta_t} \epsilon, \: \epsilon \sim \mathcal{N}(0, I) \\
&= \sqrt{1 - \beta_t} (\sqrt{1 - \beta_{t-1}} x_{t-2} + \sqrt{\beta_{t-1}} \epsilon') + \sqrt{\beta} \epsilon \\
&= \sqrt{1 - \beta_t}\sqrt{1 - \beta_{t-1}} x_{t-2} + \sqrt{1 - \beta_t}\sqrt{\beta_{t-1}} \epsilon' + \sqrt{\beta} \epsilon \\
&= \sqrt{1 - \beta_t}\sqrt{1 - \beta_{t-1}} x_{t-2} +
\sqrt{1 - (1 - \beta_t)(1 - \beta_{t-1})} \epsilon \\
&= \cdots \\
&= (\prod_{s=1}^t \sqrt{1 - \beta_s}) x_0 + \sqrt{1 - \prod_{s=1}^t (1 - \beta_s)} \epsilon,
\end{array}
\]

\[q(x_t|x_0) = \mathcal{N}(x_t|\sqrt{\bar{\alpha}_t}x_0, (1 - \bar{\alpha}_t)I), \: \bar{\alpha}_t := \prod_{s=1}^t \alpha_s, \alpha_s := 1 - \beta_s.
\]

对于后验分布\(q(x_{t-1}|x_t, x_0)\), 我们有

\[\begin{array}{ll}
q(x_{t-1}|x_t, x_0)
&= \frac{q(x_t|x_{t-1})q(x_{t-1}|x_0)}{q(x_t|x_0)} \\
&\propto q(x_t|x_{t-1})q(x_{t-1}|x_0) \\
&\propto \exp\Bigg\{-\frac{1}{2 (1 - \bar{\alpha}_{t-1})\beta_t} \bigg[(1 - \bar{\alpha}_{t-1}) \|x_t - \sqrt{1 - \beta_t} x_{t-1}\|^2 + \beta_t \|x_{t-1} - \sqrt{\bar{\alpha}_{t-1}}x_0\|^2 \bigg]\Bigg\} \\
&\propto \exp\Bigg\{-\frac{1}{2 (1 - \bar{\alpha}_{t-1})\beta_t} \bigg[(1 - \bar{\alpha}_t)\|x_{t-1}\|^2 - 2(1 - \bar{\alpha}_{t-1}) \sqrt{\alpha_t} x_t^Tx_{t-1} - 2 \sqrt{\bar{\alpha}_{t-1}} \beta_t \bigg]\Bigg\} \\
\end{array}
\]

所以

\[q(x_{t-1}|x_t, x_0) \sim \mathcal{N}(x_{t-1}|\tilde{u}_t(x_t, x_0), \tilde{\beta}_t I),
\]

其中

\[\tilde{u}_t(x_t,x_0) := \frac{\sqrt{\bar{\alpha}_{t-1}}\beta_t}{1 - \bar{\alpha}_t} x_0 + \frac{\sqrt{\alpha_t}(1 - \bar{\alpha}_{t-1})}{1 - \bar{\alpha}_t} x_t,
\]
\[\tilde{\beta}_t = \frac{1 - \bar{\alpha}_{t-1}}{1 - \bar{\alpha}_t} \beta_t.
\]

\(L_{t}\)

\(L_T\)与\(\theta\)无关, 舍去.

作者假设\(\Sigma_{\theta}(x_t, t) = \sigma_t^2 I\)为非训练的参数, 其中

\[\sigma_t^2 = \beta_t | \tilde{\beta}_t,
\]

分别为\(x_0 \sim \mathcal{N}(0, I)\)和\(x_0\)为固定值时, 期望下KL散度的最优参数(作者说在实验中二者差不多).

\[L_{t} = \frac{1}{2 \sigma^2_t} \| \mu_{\theta}(x_t, t) - \tilde{u}_t(x_t, x_0)\|^2 +C, \quad t = 1,2,\cdots, T-1.
\]

\[x_t = \sqrt{\bar{\alpha}_t} x_0 + \sqrt{1 - \bar{\alpha}_t} \epsilon \Rightarrow x_0 = \frac{1}{\sqrt{\bar{\alpha}_t}}x_t - \frac{\sqrt{1 - \bar{\alpha}_t} }{\sqrt{\bar{\alpha}_t}} \epsilon.
\]

所以

\[\begin{array}{ll}
\mathbb{E}_q [L_{t-1} - C]
&= \mathbb{E}_{x_0, \epsilon} \bigg\{
\frac{1}{2 \sigma_t^2} \| \mu_{\theta}(x_t, t) - \tilde{u}_t\big( x_t, (\frac{1}{\sqrt{\bar{\alpha}_t}}x_t - \frac{\sqrt{1 - \bar{\alpha}_t} }{\sqrt{\bar{\alpha}_t}} \epsilon) \big)\|^2 \bigg\} \\
&= \mathbb{E}_{x_0, \epsilon} \bigg\{
\frac{1}{2 \sigma^2_t} \| \mu_{\theta}(x_t, t) -
\frac{1}{\sqrt{\alpha_t}} \big( x_t - \frac{\beta_t}{\sqrt{1 - \bar{\alpha}_t}} \epsilon \big) \bigg\} \\
\end{array}
\]

注: 上式子中\(x_t\)由\(x_0, \epsilon\)决定, 实际上\(x_t = x_t(x_0, \epsilon)\), 故期望实际上是对\(x_t\)求期望.

既然如此, 我们不妨直接参数化\(\mu_{\theta}\)为

\[\mu_{\theta}(x_t, t):=
\frac{1}{\sqrt{\alpha_t}} \big( x_t - \frac{\beta_t}{\sqrt{1 - \bar{\alpha}_t}} \epsilon_{\theta}(x_t, t) \big),
\]

即直接建模残差\(\epsilon\).

此时损失可简化为:

\[\mathbb{E}_{x_0, \epsilon} \bigg\{
\frac{\beta_t^2}{2\sigma_t^2 \alpha_t (1 - \bar{\alpha}_t)} \|\epsilon_{\theta}(\sqrt{\bar{\alpha}_t}x_0 + \sqrt{1 - \bar{\alpha}_t}\epsilon, t) - \epsilon\|^2
\bigg\}
\]

这个实际上时denoising score matching.

类似地, 从\(p_{\theta}(x_{t-1}|x_t)\)采样则为:

\[x_{t-1} =
\frac{1}{\sqrt{\alpha_t}} \big( x_t - \frac{\beta_t}{\sqrt{1 - \bar{\alpha}_t}} \epsilon_{\theta}(x_t, t) \big) + \sigma_t z, \: z \sim \mathcal{N}(0, I),
\]

这是Langevin dynamic的形式(步长和权重有点变化)

注: 这部分见here.

\(L_0\)

最后我们要处理\(L_0\), 这里作者假设\(x_0|x_1\)满足一个离散分布, 首先图片取值于\(\{0, 1, 2, \cdots, 255\}\), 并标准化至\([-1, 1]\). 假设

\[p_{\theta}(x_0|x_1) = \prod_{i=1}^D \int_{\delta_{-}(x_0^i)}^{\delta_+(x_0^i) } \mathcal{N}(x; \mu_{\theta}^i(x_1, 1), \sigma_1^2) \mathrm{d}x, \\
\delta_+(x) =
\left \{
\begin{array}{ll}
+\infty & \text{if } x = 1, \\
x + \frac{1}{255} & \text{if } x < 1.
\end{array}
\right .
\delta_- (x)
\left \{
\begin{array}{ll}
-\infty & \text{if } x = -1, \\
x - \frac{1}{255} & \text{if } x > -1.
\end{array}
\right .
\]

实际上就是将普通的正态分布划分为:

\[(-\infty, -1 + 1/255], (-1 + 1 / 255, -1 + 3/255], \cdots, (1 - 3/255, 1 - 1/255], (1 - 1 / 255, +\infty)
\]

各取值落在其中之一.

在实际代码编写中, 会遇到高斯函数密度函数估计的问题(直接求是做不到的), 作者选择用下列的估计方式:

\[\Phi(x) \approx \frac{1}{2} \Bigg\{1 + \tanh \bigg(\sqrt{2/\pi} (1 + 0.044715 x^2) \bigg) \Bigg\}.
\]

这样梯度也就能够回传了.

注: 该估计属于Page.

最后的算法

注: \(t=1\)对应\(L_0\), \(t=2,\cdots, T\)对应\(L_{1}, \cdots, L_{T-1}\).

注: 对于\(L_t\)作者省略了开始的系数, 这反而是一种加权.

作者在实际中是采样损失用以训练的.

细节

注意到, 作者的\(\epsilon_{\theta}(\cdot, t)\)是有显示强调\(t\), 作者在实验中是通过attention中的位置编码实现的, 假设位置编码为\(P\):

  1. $ t = \text{Linear}(\text{ACT}(\text{Linear}(t * P)))$, 即通过两层的MLP来转换得到time_steps;
  2. 作者用的是U-Net结构, 在每个residual 模块中:
\[x += \text{Linear}(\text{ACT}(t)).
\]
参数
\(T\) 1000
\(\beta_t\) \([0.0001, 0.02]\), 线性增长\(1,2,\cdots, T\).
backbone U-Net

注: 作者在实现中还用到了EMA等技巧.

代码

原文代码

lucidrains-denoising-diffusion-pytorch

Denoising Diffusion Probabilistic Models (DDPM)的更多相关文章

  1. (转) RNN models for image generation

    RNN models for image generation MARCH 3, 2017   Today we’re looking at the remaining papers from the ...

  2. {ICIP2014}{收录论文列表}

    This article come from HEREARS-L1: Learning Tuesday 10:30–12:30; Oral Session; Room: Leonard de Vinc ...

  3. CVPR 2015 papers

    CVPR2015 Papers震撼来袭! CVPR 2015的文章可以下载了,如果链接无法下载,可以在Google上通过搜索paper名字下载(友情提示:可以使用filetype:pdf命令). Go ...

  4. ICLR 2014 International Conference on Learning Representations深度学习论文papers

    ICLR 2014 International Conference on Learning Representations Apr 14 - 16, 2014, Banff, Canada Work ...

  5. cvpr2015papers

    @http://www-cs-faculty.stanford.edu/people/karpathy/cvpr2015papers/ CVPR 2015 papers (in nicer forma ...

  6. Official Program for CVPR 2015

    From:  http://www.pamitc.org/cvpr15/program.php Official Program for CVPR 2015 Monday, June 8 8:30am ...

  7. Machine and Deep Learning with Python

    Machine and Deep Learning with Python Education Tutorials and courses Supervised learning superstiti ...

  8. Notes(一)

    Numerous experimental measurements in spatially complex systems have revealed anomalous diffusion in ...

  9. 斯坦福CS课程列表

    http://exploredegrees.stanford.edu/coursedescriptions/cs/ CS 101. Introduction to Computing Principl ...

随机推荐

  1. [学习总结]1、View的scrollTo 和 scrollBy 方法使用说明和区别

    参考资料:http://blog.csdn.net/vipzjyno1/article/details/24577023 非常感谢这个兄弟! 先查看这2个方法的源码: scrollTo: 1 /** ...

  2. JDK1.8新特性(一): 接口的默认方法default

    前言 今天在学习mysql分区优化时,发现一个博客专家大神,对其发布的文章简单学习一下: 一:简介 我们通常所说的接口的作用是用于定义一套标准.约束.规范等,接口中的方法只声明方法的签名,不提供相应的 ...

  3. shell脚本实现网站日志分析统计

    如何用shell脚本分析与统计每天的访问日志,并发送到电子邮箱,以方便每天了解网站情况.今天脚本小编为大家介绍一款不错的shell脚本,可以实现如上功能. 本脚本统计了:1.总访问量2.总带宽3.独立 ...

  4. Spring MVC入门(二)—— URI Builder模式

    URI Builder Spring MVC作为一个web层框架,避免不了处理URI.URL等和HTTP协议相关的元素,因此它提供了非常好用.功能强大的URI Builder模式来完成,这就是本文重点 ...

  5. springboot中如何向redis缓存中存入数据

    package com.hope;import com.fasterxml.jackson.core.JsonProcessingException;import com.fasterxml.jack ...

  6. vue-cli4结合element-ui异常解决(前端小白,文摘取自网络)

    1:将vue-cli4版本退回到vue-cli3 2:使用element-plus 替换 element-ui 传送门 => https://element-plus.gitee.io/#/zh ...

  7. ciscn_2019_c_1 1

    步骤: 先checksec,看一下开启了什么保护 可以看到开启了nx保护,然后把程序放入ida里面,观察程序代码 先shift+f12观察是否有system和binsh函数 发现没有system和bi ...

  8. [BUUCTF]PWN——picoctf_2018_buffer overflow 1/2

    picoctf_2018_buffer overflow 1 附件 步骤: 例行检查,32位程序,没开保护 本地运行一下程序,看看程序大概的执行情况 32位ida载入,习惯性的检查程序里的字符串,发现 ...

  9. CF999A Mishka and Contest 题解

    Content 能力值为 \(k\) 的小 M 参加一次考试,考试一共有 \(n\) 道题目,每道题目的难度为 \(a_i\).小 M 会选择两头中的一道难度不超过他的能力值题目去做,每做完一道,这道 ...

  10. linux安装软件系列之yum安装

    自动搜索最快镜像插件: yum install yum-fastestmirror 安装yum图形窗口插件: yum install yumex 1.安装 yum install 全部安装 yum i ...