\iffalse \begin{figure} \begin{floatrow}[3] \ffigbox{ \caption{Vectorisation comparison for batch learning (batch size 100, samples of 10 time steps)} \label{fig:batchcmp}}{\includegraphics[width=4.7cm,height=4.5cm]{picture//batchcomp_bold2}} \ffigbox{ \caption{Batch learning with Gradient Descent} \label{fig:batch} }{\includegraphics[width=4.7cm,height=4.5cm]{picture//batchsize_bold2}} \ffigbox{\caption{Comparison with Optimization Algorithms} \label{fig:lbfgs}} {\includegraphics[width= 4.7cm,height=4.5cm]{picture//lbfgs2_bold}} \end{floatrow} \end{figure} \fi \begin{figure*}\vspace{-.2cm} \begin{centering} \subfigure[]{\includegraphics[width=4.6cm,height=3.5cm]{picture//batchsize_bold2}\label{fig:batch}} \subfigure[]{\includegraphics[width=4.6cm,height=3.5cm]{picture//batchcomp_bold2}\label{fig:batchcmp}} \subfigure[]{\includegraphics[width= 4.6cm,height=3.5cm]{picture//lbfgs2_bold}\label{fig:lbfgs}} \end{centering}\vspace{-.2cm} \caption{(a) batch learning with gradient descent; (b) vectorization comparison for (mini-)batch learning, where the batch size is 100 and samples of 10 time steps; and (c) comparison with optimization algorithms.} \end{figure*}