Try   HackMD

Diffusion Models | Paper Explanation | Math Explained影片講解

Diffusion Models | Paper Explanation | Math Explained
Notation:
疊加x次雜訊的結果

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

為影像添加雜訊的Forward Process

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

使用神經網路訓練參數來移除雜訊的Reverse Process

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

一步Forward的細節 Bt為超參數

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

原始文章中Bt的數值為由小到大的線性數字(這樣當t趨近無窮 下面的式子會接近標準分布)

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

一個之後會用到的改寫用的符號

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

將式子改寫後 可看出從x0推到xt不在需要t步 因為最後的結果只跟x0以及alpha(t)有關

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

為影像減少雜訊的 Reverse Process 基本上就是希望機器要根據圖片 猜測這個分布的平均跟標準差

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

等等補

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →

Last changed by 
 
每日關心專題進度
專題
0
134

Read more

討論紀錄

<input class="task-list-item-checkbox" disabled="" type="checkbox"> 針對特定的圖片部分 觀察CLIP score有無變化

Feb 20, 2024
研究文章

擴散模型在圖像風格轉換中的表現一直都很出色，但最大的問題是模型本身速度不快，且擴散模型的隨機性也影響了產出的內容。大部分現有的方法需要對擴散模型進行微調，或者用額外的神經網絡。而我們使用了一種不需要額外訓練，直接使用額外的loss function來試圖將預訓練擴散模型的輸出導向到想要的方向。通過這種方法來提高使用者建構的速度，而不用花太多時間來微調擴散模型，並與其他現有不同的圖像風格轉換方法來做比較。

Dec 13, 2023
論文架構

作法A

Dec 12, 2023
風格辨識

CNN

Oct 15, 2023
Read more from 每日關心專題進度
Published on HackMD