header-includes:
\usepackage[]{algorithm2e}

# Denoising Diffusion Models

## Forward (diffusion) process

### Algorithm 1: Diffusion Process
Python-styled pseudo code for diffusion process
### Noise Scheduling

**Linear noise scheduling**

**Cosine noise scheduling**

---
Input: $x \in \mathcal{R}^{N \times H \times W}$, the noise schedule $$

---

| Column 1 | Column 2 | Column 3 |
| -------- | -------- | -------- |
|

\begin{align*}
q(x_{1:T}|x_0) &= \frac{q(x,z_{1:T})}{q(x)} \\
& = \frac{q(x,z_1)}{q(x)}\frac{q(z_1, z_2)}{q(z_1)}...\frac{q(z_{T-1},z_T)}{q(z_{T-1})} \\
& = q(z_1|x)q(z_2|z_1)...q(z_T|z_{T-1}) \\
q(x_{1:T}|x_0) &= \prod_{t=1}^{T}q(z_{t}|z_{t-1})
\end{align*}

| asfasfasfasf | Text |

## Reverse process

## Three Equivalent Forms

## Classifier guidance

## Classifier-free guidance
*