基本用法
步骤1:导入需要的算法包。
\usepackage[ruled,vlined]{algorithm2e}
步骤2:具体算法。
\begin{algorithm}[t]
\caption{Framework of Meta-GNN.}
\label{alg:algorithm1}
\KwIn{Distribution over mete-training tasks: $p(\mathcal{T})$; Meta-testing tasks: $\mathcal{T}_{mt}$; Task-learning rate: $\alpha_{1}$; Meta-learning rate: $\alpha_{2}$.}
\KwOut{Labels of nodes in query set of $\mathcal{T}_{mt}$.}
\BlankLine
Initialize $\bm{\theta}$ randomly;
\While{\textnormal{not converged}}{
Sample batch of meta-training tasks $\mathcal{T}_{i} \sim p(\mathcal{T})$;
\ForEach{task in $\mathcal{T}_{i}$}{
Evaluate $\mathcal{L}_{\mathcal{T}_{i}}\left(f_{\bm{\theta}}\right)$ using $\mathcal{S}_{i}$;
Compute adapted parameters $\bm{{\theta}^{\prime}_{i}}$;
Evaluate $\mathcal{L}_{\mathcal{T}_{i}}\left(f_{\bm{{\theta}_{i}^{\prime}}}\right)$ using $\mathcal{Q}_{i}$;
}
Update $\bm{\theta}$ by;
}
Compute adapted parameters $\bm{{\theta}^{\prime}_{mt}}$ using support set of $\mathcal{T}_{mt}$;
Predict labels of nodes in query set of $\mathcal{T}_{mt}$ using model $f_{\bm{{\theta}_{mt}^{\prime}}}$.
\end{algorithm}
最后的效果图:
算法的标题
\caption{Framework of Meta-GNN.}
算法的标签,方便在文章中引用,例如
\label{alg:algorithm1}
Algorithm~\ref{alg:algorithm1}
算法的输入
\KwIn{Distribution over mete-training tasks: $p(\mathcal{T})$; Meta-testing tasks: $\mathcal{T}_{mt}$; Task-learning rate: $\alpha_{1}$; Meta-learning rate: $\alpha_{2}$.}
算法的输出。
\KwOut{Labels of nodes in query set of $\mathcal{T}_{mt}$.}
备注:
(1).如果想要给算法的步骤加上序号:
这样引入包:
\usepackage[ruled,linesnumbered]{algorithm2e}
效果如下:
(2). 某些步骤加颜色:
首先引入包:
\usepackage{color}
然后修改其中一句
{\color{red}Evaluate $\mathcal{L}_{\mathcal{T}_{i}}\left(f_{\bm{{\theta}_{i}^{\prime}}}\right)$ using $\mathcal{Q}_{i}$;}
实际效果: