Improve article 2409.10821

[rafaelcp]

Exact location of issue
Authors and Affiliations
Table 1
Figure 3

Problem details
Elements that should be side-by-side behave weirdly, like misalignment, misplacing captions or changing image size.

Desktop (please complete the following information)

• OS: Windows
• Browser Chrome

[dginev]

Thank you for the report, these are three good tests for the different CSS flexbox heuristics in use.

• Authors

\author{Rafael C. Pinto$^1$, Anderson R. Tavares$^2$ \\
    \small \texttt{rafael.pinto@canoas.ifrs.edu.br$^1$, artavares@inf.ufrgs.br$^2$}\\
    \small Federal Institute of Rio Grande do Sul (IFRS)$^1$ \\
    \small Canoas, Brazil$^1$ \\
    \small Institute of Informatics, Federal University of Rio Grande do Sul (UFRGS)$^2$ \\
    \small Porto Alegre, Brazil$^2$ \\
}

• Table

\begin{figure}[ht]
\centering
\begin{minipage}{0.45\textwidth}
    \centering
    \includegraphics[width=\textwidth]{img/xor-prelu.pdf}
    \caption{PReLU solution to the XOR problem with inputs in $\{0,1\}$. Connection weights are interchangeable. Halve connection weights for inputs in $\{-1,1\}$.}
    \label{fig:xor-prelu}
\end{minipage}
\hspace{0.05\textwidth}%
\begin{minipage}{0.45\textwidth}
    \centering
    \vspace{0.01\textwidth}%    
    \begin{tabular}{|c|c|c|c|}
    \hline
    $x_1$ & $x_2$ & $x_1 \oplus x_2$ & $|x_1 - x_2|$ \\
    \hline
    0 & 0 & 0 & 0 \\
    0 & 1 & 1 & 1 \\
    1 & 0 & 1 & 1 \\
    1 & 1 & 0 & 0 \\
    \hline
    \end{tabular}
    \captionof{table}{Truth table for XOR operation and $|x_1 - x_2|$, which is equivalent to the PReLU activation for $a = -1$.}
    \label{tab:xor_truth_table}
\end{minipage}%
\end{figure}

• Figure

\begin{figure}[htb]
    \centering
    \begin{minipage}[t]{0.48\textwidth}
        \centering
        \includegraphics[width=\textwidth]{img/learning_rate_success.pdf}
        \caption{Success rate vs learning rate (100 trials, 300 epochs). PReLU shows the widest range of acceptable learning rates for the Adam optimizer in the XOR problem. In a separate experiment, PReLU with a bias shows similar behavior to the other models when close to learning rate 1.}
        \label{fig:accuracy_lr_plot}
    \end{minipage}
    \hfill
    \begin{minipage}[t]{0.48\textwidth}
        \centering
        \includegraphics[width=\textwidth,height=130px]{img/runtime_boxplot.pdf}
        \caption{Runtime distribution for each model for a single trial (300 epochs). PReLU and GCU (3 learnable parameters each) are significantly faster than the MLP solution (8 learnable parameters) and have similar runtime.}
        \label{fig:runtime}
    \end{minipage}
\end{figure}