/
paper_plots.tex
28 lines (26 loc) · 1.3 KB
/
paper_plots.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
% TeX file to generate paper plots after the experiments have been run.
\documentclass{article}
\usepackage{graphicx}
\usepackage[utf8]{inputenc}
\usepackage{pgfplots}
\begin{document}
\begin{figure*}[t!]
\centering
\begin{tabular}{ccc}
\resizebox{.31\linewidth}{!}{\input{paper_experiments/exact_vs_inexact_prox_mlp_plot.tex}} &
\resizebox{.31\linewidth}{!}{\input{paper_experiments/inexact_prox_vs_sgd_mlp_plot.tex}} &
\resizebox{.31\linewidth}{!}{\input{paper_experiments/exact_vs_inexact_prox_mlp_plot_val_acc.tex}}
\end{tabular}
\caption{Exact and inexact solvers for ProxProp compared with BackProp for an MLP on CIFAR-10.}
\end{figure*}
\begin{figure*}[t!]
\centering
\begin{tabular}{cc}
\resizebox{.47\linewidth}{!}{\input{paper_experiments/proxprop_vs_sgd_adam_convnet_epochs_plot.tex}} &
\resizebox{.47\linewidth}{!}{\input{paper_experiments/proxprop_vs_sgd_adam_convnet_time_plot.tex}} \\
\resizebox{.47\linewidth}{!}{\input{paper_experiments/proxprop_vs_sgd_adam_convnet_epochs_plot_val.tex}} &
\resizebox{.47\linewidth}{!}{\input{paper_experiments/proxprop_vs_sgd_adam_convnet_time_plot_val.tex}} \\
\end{tabular}
\caption{ProxProp as a first-order oracle in combination with the Adam optimizer for a convolutional neural network on CIFAR-10.}
\end{figure*}
\end{document}