Seminar_in_AI/presentation/main.tex

\documentclass[usenames,dvipsnames]{beamer}
%----------------------------------------------------------------------------------------
%	Struktur und Pointer Referat
%   20.04.2020
%----------------------------------------------------------------------------------------
\usetheme{focus}

\usepackage[utf8]{inputenc}

\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{hyperref}
\usepackage{graphicx}
\usepackage{listings}
\usepackage{xcolor}

% Farbdefinitionen
\definecolor{backgroundcoloreq}{RGB}{180,140,0}
\definecolor{codegreen}{rgb}{0,0.6,0}
\definecolor{codegray}{rgb}{0.5,0.5,0.5}
\definecolor{codepurple}{rgb}{0.58,0,0.82}
\definecolor{codeorange}{RGB}{190,100,0}


\lstset{
    language=C,
    basicstyle=\ttfamily,
    numbers=left,
    numberstyle=\tiny,
    tabsize=4,
    columns=fixed,
    showstringspaces=false,
    showtabs=false,
    breaklines=true,
    keepspaces,
    morekeywords={std},
    keywordstyle=\color{blue}\ttfamily,
    stringstyle=\color{red}\ttfamily,
    commentstyle=\color{OliveGreen!85}\ttfamily,
    numberstyle=\tiny\color{codegray},
    basicstyle=\footnotesize\selectfont\ttfamily,
% keyword highlighting
    classoffset=1, % starting new class
    otherkeywords={>,<,.,;,-,+,!,=,~,:,[,],NULL,&},
    morekeywords={>,<,.,;,-,+,!,=,~,:,[,],NULL,&},
    keywordstyle=\color{codeorange},
    classoffset=0
}

%----------------------------------------------------------------------------------------
%	 TITLE SLIDE
%----------------------------------------------------------------------------------------

\title{Cross-Model Pseudo-Labeling}

\subtitle{for Semi-Supervised Action Recognition}

\author{Lukas Heiligenbrunner}

\date{\today}

%------------------------------------------------

\begin{document}

%------------------------------------------------

    \begin{frame}
        \maketitle
    \end{frame}


%----------------------------------------------------------------------------------------
%	 SECTION 1
%----------------------------------------------------------------------------------------
% todo pic of action

    \section{The goal}
    \begin{frame}{The goal}
        \begin{itemize}
            \item train model
            \item recognize action of person
            \item from video [$\approx$10sec]
            \item eg.:
            \begin{itemize}
                \item brushing hair
                \item riding bike
                \item dancing
                \item playing violin
            \end{itemize}
            \item as generic as possible
        \end{itemize}
    \end{frame}

%----------------------------------------------------------------------------------------
%	 SECTION 2
%----------------------------------------------------------------------------------------


    \section{The Problem} % Section title slide, unnumbered

%------------------------------------------------

    \begin{frame}{Missing Labels}
        \begin{itemize}
            \item Supervised action recoginition
            \begin{itemize}
                \item lots of labeled samples necessary
                \item videos
            \end{itemize}
            \item Labeling Samples very expensive
            \begin{itemize}
                \item Avoid!
            \end{itemize}
            \item Tremendous amount of unlabled data
            \begin{itemize}
                \item YouTube
            \end{itemize}
            \item using semi-supervised learning might be benefitial
        \end{itemize}
    \end{frame}

%------------------------------------------------

    \begin{frame}{What's all about Semi supervised?}
        \begin{itemize}
            \item Supervised learning
            \begin{itemize}
                \item Data samples
                \item Target labels
                \item Each sample is associated to target label
            \end{itemize}
            \item Unsupervised learning
            \begin{itemize}
                \item Data samples
                \item target is to find patterns in data
                \item without supervision
            \end{itemize}
            \item Semi-Supervised learning
            \begin{itemize}
                \item combination of both
                \item have labeled \& unlabeled data
                \item labeled data guides learning process
                \item unlabled helps to gain additional information
                \item goal is performance improvement
            \end{itemize}
        \end{itemize}
    \end{frame}

%------------------------------------------------

    \begin{frame}[allowframebreaks]{What's already been done}
        \begin{itemize}
            \item Pseudo-labeling
            \item Train model on labeled data
            \begin{itemize}
                \item Eg. 1\% of data labeled
            \end{itemize}
            \item Confidence of prediction
            \item If high enough
            \item Use to predict unlabeled data
        \end{itemize}

        \framebreak
        \begin{itemize}
            \item quantity and quality of pseudo-labels
            \item significant impact on main model accuracy!
            \item we want to improve pseudo-label framework as much as possible
        \end{itemize}
    \end{frame}

%----------------------------------------------------------------------------------------
%	 SECTION 2
%----------------------------------------------------------------------------------------

    \section{Cross-Model Pseudo-Labeling}

    \begin{frame}[allowframebreaks]{Papers approach}
        \begin{itemize}
            \item Based on complementary-representations of model
            \item Models of different size
            \item Different structural-bias $\rightarrow$ different category-wise performance
            \item Small model
            \begin{itemize}
                \item lower capacity
                \item better captures temporal dynamics in recognizing actions
            \end{itemize}
            \item Large model
            \begin{itemize}
                \item better learns spatial semantics
                \item to distinguish different action instances
            \end{itemize}
        \end{itemize}

        \framebreak

        \begin{itemize}
            \item Cross-Model Pseudo-Labeling
            \item Primary backbone
            \item Supplemented by lightweight auxiliary network
            \begin{itemize}
                      \item Different structure
                      \item Fewer channels
            \end{itemize}
            \item Different representation of data complements primary backbone
        \end{itemize}
    \end{frame}


    \begin{frame}{Performance glance}
        todo the pic of the performance graph
    \end{frame}



    % ---  THE END

    \begin{frame}[focus]
        Thanks for your Attention!
    \end{frame}

%----------------------------------------------------------------------------------------
%	 CLOSING/SUPPLEMENTARY SLIDES
%----------------------------------------------------------------------------------------

    \appendix

    \begin{frame}{Sources}
        \nocite{*} % Display all references regardless of if they were cited
        \bibliography{sources}
        \bibliographystyle{plain}
    \end{frame}

\end{document}