\documentclass{beamer}

\usepackage{ctex}
\usepackage{braket}
\usepackage{graphicx}
\graphicspath{{pics/}}
\usepackage{subfigure}  %插入子图用
\setbeamertemplate{caption}[numbered]  %让插入的图片自动编号(beamer默认不编号)
\usefonttheme{serif}   %在公式中使用使用标准的Latex字体(有衬线的字体)，区分小的L和大写的I
%beamer中默认使用Sans Serif字体，即没有衬线的字体
\renewcommand{\thefootnote}{}    %取消脚注的编号，在后面花括号内添加其他符号也可改变脚注符号
\usepackage{caption}
\usepackage[english]{babel} % 将时间改成英
\usepackage{calligra}
\setbeamertemplate{navigation symbols}

% 修改beamer脚注的默认设置
\usepackage{etoolbox}  % 引入etoolbox以方便进行宏的修补
\makeatletter
\defbeamertemplate*{footnote}{nonumber}{
	\parindent 1em\noindent% 
	\raggedright
	\insertfootnotetext\par
}
\setbeamertemplate{footnote}[nonumber]  % 应用自定义的脚注模板
\makeatother

% 幻灯片字体和主题
\usepackage{times}
\usetheme{CambridgeUS}
\usecolortheme{dolphin}

% HITsz校徽，badge文件为校徽
\pgfdeclareimage[height=0.6cm]{badge}{badge}
\logo{\pgfuseimage{badge}}

% 目录设置
\AtBeginSection[]{
	\begin{frame}
		\tableofcontents[currentsection]
	\end{frame}
}

% 首页信息
\title[JSAC]
{Sora: A Review on Background, Technology, Limitations, and Opportunities of Large Vision Models}

\author[San Zhang ]
{\fontsize{8}{10}\selectfont
	San Zhang\inst{1}, Si Li\inst{2}, and Wu Wang\inst{3}
	\\[10pt]
	\href{mailto:zhangsan@hit.edu.cn}
	{Email: \textit{zhangsan@hit.edu.cn}}}

\institute[HITsz]
{\fontsize{6}{10}\selectfont
	\inst{1, 2, 3}School of Electronics and Information Engineering\\
	Harbin Institute of Technology, Shenzhen}

% \date[\today]
\date[May 18, 2024]
{\fontsize{8}{10}
	\selectfont May 18, 2024}

% 正文
\begin{document}
	\logo{}
	\begin{frame}
		\titlepage
		\begin{center}
			\vspace{-15pt}
			\includegraphics[width=0.3\linewidth]{Badge.png}
		\end{center}
	\end{frame}
	
	% 目录
	\logo{\pgfuseimage{badge}}
	\begin{frame}
		\frametitle{Contents}
		\tableofcontents
	\end{frame}
	\logo{}
	
	
	
	
	
	% Section 1
	\section[Introduction]{Introduction}
	\begin{frame}
		\frametitle{The Development of AI-Generated Content (AIGC)}
		\begin{figure}[t]
			\centering
			\includegraphics[width=3.7in]{Figures/DevelopmentOfAIGC}
			\vspace{0mm}
			\captionsetup{font=scriptsize}
			\caption{The development of AIGC and mobile edge computing network [1].}
			\vspace{0mm}
		\end{figure}
		\footnotetext{\noindent \tiny [1] M. Xu, \emph{et al.}, ``Unleashing the power of edge-cloud generative AI in mobile networks: A survey of AIGC services,'' \emph{IEEE Commun. Surv. Tutor.}, Early Access, 2024.}
	\end{frame}


	\begin{frame}
		\frametitle{Diffusion Model}
		\begin{figure}[t]
			\centering
			\includegraphics[width=3.8in]{Figures/DiffusionModel}
			\vspace{0mm}
			\captionsetup{font=scriptsize}
			\caption{An illustration of diffusion model [2].}
			\vspace{-2mm}
		\end{figure}
		\begin{itemize}
			\item With Gaussian noise as input, the well-trained diffusion model is capable of generating new contents through a series of inference steps.
			\item The quality of generated images gets progressively better as the number of inference steps increases.
		\end{itemize}
		\footnote{\noindent \tiny{[2] H. Du, \emph{et al.}, ``Enhancing deep reinforcement learning: A tutorial on generative diffusion models in network optimization," \emph{arXiv preprint arXiv:2308.05384}, 2023.}}
	\end{frame}
	
	
	\begin{frame}
	\frametitle{AIGC in Mobile Edge Computing (MEC)}
		\begin{figure}[t]
			\centering
			\includegraphics[width=2.6in]{Figures/AIGC+MEC}
			\vspace{0mm}
			\captionsetup{font=scriptsize}
			\caption{An overview of a mobile AIGC network [1].}
			\vspace{0mm}
		\end{figure}
		\begin{itemize}
			\vspace{-3mm}
			\item AIGC models can be deployed at edge servers.
		\end{itemize}
		\footnotetext{\noindent \tiny [1] M. Xu, \emph{et al.}, ``Unleashing the power of edge-cloud generative AI in mobile networks: A survey of AIGC services,'' \emph{IEEE Commun. Surv. Tutor.}, Early Access, 2024.}
	\end{frame}

	
	
	
	
	% Section 2
	\section[System Model]{System Model}
	\subsection{Network Model}
	\begin{frame}
		\frametitle{MEC Network}
	\end{frame}
	
	
	\subsection{Task Processing Model}
	\begin{frame}
		\frametitle{Local Processing Model}
	\end{frame}
	
	
	\begin{frame}
		\frametitle{Home BS Processing Model}
	\end{frame}
	
	
	\begin{frame}
		\frametitle{Neighbor BS Processing Model}
	\end{frame}
	
	
	
	
	
	% Section 3
	\section[Problem Formulation]{Problem Formulation}
	\begin{frame}
		\frametitle{Weighted Cost}
	\end{frame}


	\begin{frame}
		\frametitle{Offloading Problem}
	\end{frame}





	% Section 4
	\section[Algorithm Design]{Algorithm Design}
	\begin{frame}
		\frametitle{Deep Reinforcement Learning based OSI Algorithm}
		\begin{itemize}
			\item State:
			$$
			\bold{s}_n^{(l)} = \left( B_n^{(l)}, q_n^{(l)}, f_n^{\text{U},(l)}, \bold{g}^{\text{B},(l)}, \bold{h}_n^{(l)} \right)
			$$
			\item Action:
			$$
			\bold{a}_n^{(l)} = \left( x_n^{(l)}, \bold{y}_{n}^{(l)}, c_n^{(l)} \right)
			$$
			\item Reward:
			$$
			r_n^{(l)} = -\sum_{n \in \mathcal{N}} \left( \omega_1 T_n^{(l)} + \omega_2 E_n^{(l)} + \omega_3 \epsilon_n^{(l)} \right)-r_n^{\text{P},(l)}
			$$
		\end{itemize}
	\end{frame}





	% Section 5
	\section[Simulation Results]{Simulation Results}
	\begin{frame}
	\end{frame}





	% Section 6
	\section[Conclusion]{Conclusion}
	\begin{frame}
		\begin{itemize}
			\item Conclusion 1.
			\item Conclusion 2.
			\item Conclusion 3.
			\item Conclusion 4.
		\end{itemize}
	\end{frame}


	\section{}
	\logo{\pgfuseimage{badge}}
	\begin{frame}{Acknowledgements}
		\begin{center}
			\begin{minipage}{1\textwidth}
				\setbeamercolor{mybox}{fg=white, bg=black!50!blue}
				\begin{beamercolorbox}[wd=0.70\textwidth, rounded=true, shadow=true]{mybox}
					\LARGE \centering Thanks for your listening!
				\end{beamercolorbox}
			\end{minipage}
			\\[20pt]
			Please feel free to contact us:
			\\[10pt]
			\textit{zhangsan@hit.edu.cn}
		\end{center}
	\end{frame}

\end{document}

\title[Private Learner is Online Learner] %optional # Paper Title
{An Equivalence Between Private Classification and Online Prediction
}
% \subtitle{FOCS 2020} % 可以加你要present的paper发表在哪里，比如我要present的是FOCS最佳论文
\author[Yucheng Sun, Jiaqing Xie] % (optional) % Presenter Name
{Mark Bun \inst{1} , Roi Livni \inst{2}, Shay Moran \inst{3}}

%The next block of commands puts the table of contents at the 
%beginning of each section and highlights the current section:
\AtBeginSection[]
{
  \begin{frame}
    \frametitle{Table of Contents} 
    \tableofcontents[currentsection]

  \end{frame}
 
}

\newtheorem{Proposition}{Proposition}
\newtheorem{Algorithm}{Algorithm}

\begin{frame}{Online Learning} % 每页的开始用frame

    \begin{definition}[Goal of Online Learning] % Definition 后面跟着的
        We want to minimize the regret, the number of mistakes compared to the best hypothesis in H:
        \[
        R(n) = \sum_{t = 1}^{n}1[y_t \neq \hat{y_t}] -  \min_{h^* \in H} \sum_{t = 1}^{n}1[y_t \neq h^*] 
        \] % 正常打你的公式
    \end{definition}   

    \begin{theorem}[Boundedness of R(n)] % 和上面的definition 一样
         R(n) has been proved to be bounded:
         \[
         \Omega(\sqrt{dn}) \leq R(n) \leq O(\sqrt{dn \log n})
         \]
         where d is the Littlestone Dimension
    \end{theorem}
           
\end{frame}

\begin{frame}{Standard Optimal Algorithm (SOA)}
    \begin{definition}[Realizable to non-realizable samples]
       So far samples could be shattered by H $\Leftrightarrow$ Samples are realizable. 
% ############### start your algorithm ###########################
       \begin{enumerate} 
           \item If the incoming sample $x_{t+1}, y_{t+1}$ maintains the realizability, update H by SOA.
           \item Else:  % 每一行用item
           \[
            h'(x_{t+1}) = y_{t+1} 
           \] while keeping other $h'(\cdot)$ unchanged. % 正常打公式就行。
       \end{enumerate}
% ############### end your algorithm ###########################
    \end{definition}
\end{frame}

\begin{frame}{Proof}
    \begin{itemize}
        \item Assume optimal $k^*$ and $f = f^*$, we  have $loss_D(f^*) \leq loss_D(f) = \alpha$
        \item Markov's inequality: \[
        \mathbb{P}[M_{k^*} >  2^{2^{d+2} + 1} 4^{d+1} \cdot  \lceil \frac{2^{d+2}}{\alpha} \rceil ] \leq 2^{-2^{d+2}-1}
        \] 
        \item $\underset{S \sim \hat{D}_{k^*}, T \sim D^n}{\mathbb{P}} [SOA(S  \circ T) = f^*] = \underset{S \sim D_{k^*}, T \sim D^n}{\mathbb{P}} [SOA(S  \circ T) = f^* \quad \text{and} \quad M_{k^*} < 2^{2^{d+2} + 1} 4^{d+1} \cdot  \lceil \frac{2^{d+2}}{\alpha} \rceil ] \geq 2^{-2^d -1} \geq 2^{-2^{d+2} -1}$
        \item $k = k^*$ with probability $\frac{1}{d+1}$. Finish the proof.
    \end{itemize}
\end{frame}

\documentclass{beamer}
\usepackage{包的名称}
\usetheme{主题名称}
\usecolortheme{颜色主题名称}
\usefonttheme{字体主题名称}

\begin{document}
标题、作者、单位和日期等信息
\begin{frame} 
幻灯片内容
\end{frame}
......
\begin{frame} 
幻灯片内容
\end{frame}
\end{document}

\title{Applied Microeconometrics}
\author{Jiang He}
\institute{Central China Normal University}
\date{\today}

\begin{frame} 
\frametitle{推荐使用教材} 
\begin{itemize} 
 \item 基本无害的计量经济学
 \item 基本有用的计量经济学 
 \item 因果推断实用计量方法
 \item 精通计量：从原因到结果的探寻之旅 
\end{itemize} 
\end{frame}

\begin{frame}
\frametitle{潜在结果框架}
假如我们想要考察上大学对个人收入的影响，那么对于任意个体$i$就有两种干预状态，一种是上大学($D_i=1$)，另一种是不上大学($D_i=0$)，每种状态对应一个潜在结果，上大学对应潜在结果收入$Y_{1i}$，不上大学对应潜在结果收入$Y_{0i}$。对于个体$i$，上大学对收入的因果效应是两种状态下潜在结果的比较：
\begin{align}
\tau_i=Y_{1i}-Y_{0i}
\end{align}
\end{frame}

"excalidraw.image": {
        "exportScale": 3,
        "exportWithBackground": false,
        "exportWithDarkMode": false
    }