data/tables/dev/val_tab_0019.tex

\documentclass[varwidth=598pt]{standalone}

\usepackage{booktabs}      % professional-quality tables
\usepackage{multirow}      % multi-row cells in tables
\usepackage{colortbl}      % color for tables (\cellcolor, \rowcolor)
\usepackage[table]{xcolor} % enhanced colors for tables
\usepackage{array}         % more flexible column formats (often used)
\usepackage{tabularx}      % for tables with auto-stretch columns (if used)
\usepackage{graphicx}      % for images in tables or figure/table floats
\usepackage{amssymb}       % math symbols
\usepackage{amsmath}       % math environments
\usepackage{soul}          % highlighting (used for colored text/cells)
\usepackage[normalem]{ulem} % underlining, strikethroughs
\usepackage[T1]{fontenc}   % font encoding
\usepackage[utf8]{inputenc}% input encoding (legacy, fine for pdflatex)
\usepackage{microtype}     % better text appearance
\usepackage{hyperref}      % hyperlinks
\usepackage{textcomp}      % for extra symbols
\usepackage{enumitem}      % for compact lists (if used in table notes)
\usepackage{adjustbox}
\usepackage{tabu}
\usepackage{pifont}        % http://ctan.org/pkg/pifont
\usepackage{bbding}        % \XSolidBrush
\usepackage{makecell}

\begingroup
\makeatletter
\renewcommand{\fnum@table}{}%
\long\def\@makecaption#1#2{%
  \vskip\abovecaptionskip
  \centering #2\par
  \vskip\belowcaptionskip
}
\makeatother


\begin{document}
\begin{table}
\centering
\begin{adjustbox}{max width=\textwidth}
\begin{tabular}{rccccc}
\hline
Method                    & Training Data             & K400        & MiT         & SSv2        & ActNet      \\ \hline
\fancyname                      & \begin{tabular}[c]{@{}l@{}}(e),(f),(g),(h) \end{tabular} & 81.9 / 95.2 & 41.7 / 71.0 & 68.9 / 91.6 & 87.4 / 97.3 \\ \hline \hline
Vanilla (50 ep)  & \begin{tabular}[c]{@{}l@{}}(e),(f), (g),(h) \end{tabular} & 80.1/ 94.0  & 33.4 / 60.1 & 60.8 / 89.0 & 86.5 / 97.1 \\ \hline
Vanilla (200 ep) & \begin{tabular}[c]{@{}l@{}}(e),(f), (g),(h) \end{tabular} & 80.6 / 94.7 & 35.1 / 63.9 & 56.8 / 85.3 & 86.3 / 97.2 \\ \hline
w/o Informative Loss        & \begin{tabular}[c]{@{}l@{}}(e),(f),(g),(h) \end{tabular} & 13.5 / 33.4 & 7.3 /  19.9 & 9.7 / 28.5  & 24.8 / 54.3 \\ \hline
\begin{tabular}[c]{@{}l@{}}w/o Informative Loss \\ \& w/o Projection Add\end{tabular}        & \begin{tabular}[c]{@{}l@{}}(e),(f),(g),(h) \end{tabular} & 80.4 / 94.5 & 38.7 / 68.9 & 62.6 / 89.8  & 86.5 / 97.4 \\ \hline
w/o Projection Loss         & \begin{tabular}[c]{@{}l@{}}(e),(f),(g),(h) \end{tabular} &      80.6 / 94.8	       &        39.9 / 69.2	     &  61.5 / 88.0	           &       86.9 / 97.5      \\ \hline
w/o ActNet                  & (e),(f),(g)         &       81.4 / 95.0	      &        41.3 / 70.5      &      68.7 / 91.3	      &       -     \\ \hline
\end{tabular}
\end{adjustbox}
\caption{Table 3: Ablation experiments. We investigate the effectiveness of each component of our method as well as compare to vanilla multi-dataset training method. 
``Vanilla'' means using cross entropy (CE) loss in training.
``w/o informative los'' means using CE and projection loss.
The numbers are top-1/top-5 accuracy, respectively.
Training data:
(e) Kinetics-400;
(f) SSv2;
(g) MiT;
(h) ActivityNet.}
\end{table}
\end{document}