finish first round of presentation

presentation/main.tex

@@ -5,14 +5,19 @@
 %----------------------------------------------------------------------------------------
 \usetheme{focus}
 
+
 \usepackage[utf8]{inputenc}
 
 \usepackage{booktabs}
 \usepackage{amsmath}
+\usepackage{amssymb}
+\usepackage{amsfonts}
+\usepackage{bbm}
 \usepackage{hyperref}
 \usepackage{graphicx}
 \usepackage{listings}
 \usepackage{xcolor}
+\usepackage{mathtools}
 
 % Farbdefinitionen
 \definecolor{backgroundcoloreq}{RGB}{180,140,0}
@@ -21,6 +26,9 @@
 \definecolor{codepurple}{rgb}{0.58,0,0.82}
 \definecolor{codeorange}{RGB}{190,100,0}
 
+% we wanna use default caleographic alphabet
+\DeclareMathAlphabet{\mathcal}{OMS}{cmbrs}{m}{n}
+
 
 \lstset{
     language=C,
@@ -75,20 +83,20 @@
 %----------------------------------------------------------------------------------------
 % todo pic of action
 
-    \section{The goal}
+    \section{The Goal}
     \begin{frame}{The goal}
         \begin{itemize}
-            \item train model
+            \item Train model
-            \item recognize action of person
+            \item Recognize action of person
-            \item from video [$\approx$10sec]
+            \item From video [$\approx$10sec]
-            \item eg.:
+            \item E.g.:
             \begin{itemize}
                 \item brushing hair
                 \item riding bike
                 \item dancing
                 \item playing violin
             \end{itemize}
-            \item as generic as possible
+            \item As generic as possible
         \end{itemize}
     \end{frame}
 
@@ -173,6 +181,7 @@
 %	 SECTION 2
 %----------------------------------------------------------------------------------------
 
+
     \section{Cross-Model Pseudo-Labeling}
 
     \begin{frame}[allowframebreaks]{Papers approach}
@@ -196,21 +205,138 @@
 
         \begin{itemize}
             \item Cross-Model Pseudo-Labeling
-            \item Primary backbone
+            \item Primary backbone (large model)
             \item Supplemented by lightweight auxiliary network
             \begin{itemize}
-                      \item Different structure
+                \item Different structure
-                      \item Fewer channels
+                \item Fewer channels
             \end{itemize}
             \item Different representation of data complements primary backbone
         \end{itemize}
     \end{frame}
 
-
+    \begin{frame}{Structure Visualization}
-    \begin{frame}{Performance glance}
+        \includegraphics[scale=.17]{rsc/structure}
-        todo the pic of the performance graph
     \end{frame}
 
+    \begin{frame}{Performance Perspectives}
+        \includegraphics[scale=.205]{rsc/performance_comparison}
+    \end{frame}
+
+
+    \section{Give me the math!}
+
+    \begin{frame}{Definitions}
+        \begin{itemize}
+            \item Labeled data set of size $N_l$\\
+            $\mathcal{V} = \{(v_1,y_1), \dots, (v_{N_l}, y_{N_l})\}$
+            \item Unlabeled data set of size $N_u$\\
+            $\mathcal{U} = \{u_1, \dots, u_{N_u}\}$
+            \item in general $\lvert\mathcal{U}\rvert \gg \lvert\mathcal{V}\rvert$\\
+        \end{itemize}
+    \end{frame}
+
+    \begin{frame}[allowframebreaks]{How existing method \textit{FixMatch} works}
+        \begin{itemize}
+            \item $\mathbbm{1} \coloneqq \text{Indicator Function}$
+            \item $B_u \coloneqq \text{Batchsize}$
+            \item $\mathcal{T} \coloneqq \text{Confidence Threshold}$
+            \item $F(\mathcal{T}_{\text{strong}}(u_i)) \coloneqq \text{Class distribution}$
+            \item $p_i \coloneqq F(\mathcal{T}_{\text{weak}}(u_i))$
+            \item $\hat{y}_i \coloneqq \arg \max(p_i) \coloneqq \text{Pseudo Label}$
+            \item $\mathcal{H} \coloneqq \text{Cross-entropy loss}$
+            \item $\mathcal{L}_u \coloneqq \text{Loss on the unlabeled data}$
+            \item $\mathcal{F} \coloneqq \text{Model}$
+        \end{itemize}
+        \begin{align*}
+            \mathcal{L}_u = \frac{1}{B_u} \sum_{i=1}^{B_u} \mathbbm{1}(\max(p_i) \geq \mathcal{T}) \mathcal{H}(\hat{y}_i,F(\mathcal{T}_{\text{strong}}(u_i)))
+        \end{align*}
+
+        \framebreak
+
+        \begin{itemize}
+            \item $\mathbbm{1}(\max(p_i) \geq \mathcal{T})$
+            \begin{itemize}
+                \item 'confidence-based masking'
+                \item retain label only if largest probability is above threshold
+                \item keep only 'high confidence' labels
+            \end{itemize}
+            \item $\mathcal{H}(\hat{y}_i,F(\mathcal{T}_{\text{strong}}(u_i)))$
+            \begin{itemize}
+                \item 'consistency regularization'
+                \item cross-entropy loss of strong augmented and weak augmented data
+            \end{itemize}
+        \end{itemize}
+
+    \end{frame}
+
+    \begin{frame}[allowframebreaks]{CMPL (Cross-Model Pseudo-Labeling)}
+        \begin{itemize}
+            \item $F(\cdot) \coloneqq \text{Primary backbone}$
+            \item $A(\cdot) \coloneqq \text{Auxiliary network}$
+            \item Learning on labeled data
+            \begin{align*}
+                \mathcal{L}_s^F &= \frac{1}{B_l} \sum_{i=1}^{B_l} \mathcal{H}(y_i,F(\mathcal{T}^F_{\text{standard}}(v_i)))\\
+                \mathcal{L}_s^A &= \frac{1}{B_l} \sum_{i=1}^{B_l} \mathcal{H}(y_i,A(\mathcal{T}^F_{\text{standard}}(v_i)))
+            \end{align*}
+            \item $\mathcal{T}^F_{\text{standard}}(v_i) \coloneqq \text{standard augmentations for action recognition}$
+        \end{itemize}
+
+        \framebreak
+
+        \begin{itemize}
+            \item Learning on unlabeled data
+            \begin{align*}
+                \mathcal{L}_u^F &= \frac{1}{B_u} \sum_{i=1}^{B_u} \mathbbm{1}(\max(p_i^A) \geq \mathcal{T}) \mathcal{H}(\hat{y}_i^A,F(\mathcal{T}_{\text{strong}}(u_i)))\\
+                \mathcal{L}_u^A &= \frac{1}{B_u} \sum_{i=1}^{B_u} \mathbbm{1}(\max(p_i^F) \geq \mathcal{T}) \mathcal{H}(\hat{y}_i^F,A(\mathcal{T}_{\text{strong}}(u_i)))\\
+            \end{align*}
+            \item Complete training objective
+            \begin{align*}
+                \mathcal{L} = (\mathcal{L}_s^F + \mathcal{L}_s^A) + \lambda(\mathcal{L}_u^F + \mathcal{L}_u^A)
+            \end{align*}
+            \item $\lambda \coloneqq \text{Balancing coefficient for unsupervised loss}$
+        \end{itemize}
+    \end{frame}
+
+    \section{Implementation}
+
+    \begin{frame}{Networks}
+        \begin{itemize}
+            \item Auxiliary Network
+            \begin{itemize}
+                \item sub-network of primary model
+                \item 3D-ResNet18
+                \item \textbf{3D-ResNet50x1/4}
+            \end{itemize}
+            \item Backbone network
+            \begin{itemize}
+                \item larger version of aux-net
+                \item \textbf{3D-ResNet50}
+            \end{itemize}
+        \end{itemize}
+    \end{frame}
+
+    \begin{frame}{Dataset}
+        \begin{itemize}
+            \item Kinetics-400
+            \begin{itemize}
+                \item 400 categories
+                \item 240k/20k training/validation samples
+            \end{itemize}
+            \item UCF-101
+            \begin{itemize}
+                \item 101 classes
+                \item 9.5k/4k training/validation samples
+            \end{itemize}
+            \item $\approx$10sec every video
+            \item 1\% or 10\% labeled subsets balanced sampled from distribution
+        \end{itemize}
+    \end{frame}
+
+
+    \begin{frame}{Performance Results}
+        \includegraphics[scale=.65]{rsc/results}
+    \end{frame}
 
 
     % ---  THE END

presentation/sources.bib

@@ -1,18 +1,16 @@
-@misc{structtutorialspoint,
+@InProceedings{Xu_2022_CVPR,
-    Title = {struct basics},
+    author    = {Xu, Yinghao and Wei, Fangyun and Sun, Xiao and Yang, Ceyuan and Shen, Yujun and Dai, Bo and Zhou, Bolei and Lin, Stephen},
-    howpublished = {\url{https://www.tutorialspoint.com/cprogramming/c_structures.htm}},
+    title     = {Cross-Model Pseudo-Labeling for Semi-Supervised Action Recognition},
-    note = {Aufgerufen: 2020-04}
+    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
+    month     = {June},
+    year      = {2022},
+    pages     = {2959-2968}
 }
 
-@misc{structchowto,
+@online{knuthwebsite,
-    Title = {struct basics},
+    author = "Kihyuk Sohn, David Berthelot, Chun-Liang Li",
-    howpublished = {\url{http://www.c-howto.de/tutorial/strukturierte-datentypen/strukturen/}},
+    title = "FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence",
-    note = {Aufgerufen: 2020-04}
+    url  = "https://arxiv.org/abs/2001.07685",
+    addendum = "(accessed: 20.03.2023)",
+    keywords = "FixMatch, semi-supervised"
 }
-
-@misc{pointertutorialspoint,
-    Title = {Pointer basics},
-    howpublished = {\url{https://www.tutorialspoint.com/cprogramming/c_pointers.htm}},
-    note = {Aufgerufen: 2020-04}
-}
-