Skip to content

T
Toy-MT-Introduction
Commit baf58219 by Lee
Options

Merge branch 'master' into liyanyang

parents cee12b1e 4f5d290b
正在显示 包含 45 行增加 和 31 行删除
Section06-Neural-Machine-Translation/section06.tex
...@@ -232,7 +232,7 @@ ...@@ -232,7 +232,7 @@
%%% 神经机器翻译的性能增长 %%% 神经机器翻译的性能增长
\begin{frame}{神经机器翻译的进展(续)} \begin{frame}{神经机器翻译的进展(续)}
\begin{itemize} \begin{itemize}
\item 神经机器翻译在大部分场景下已经超越统计机器翻译！ \item 神经机器翻译在很多场景下已经超越统计机器翻译
{ {
\footnotesize \footnotesize
\begin{center} \begin{center}
...@@ -257,7 +257,7 @@ ...@@ -257,7 +257,7 @@
\end{tabular} \end{tabular}
\end{center} \end{center}
} }
\item 微软报道在部分场景下机器翻译质量已经超越人类！ \item 微软的报道：在部分场景下机器翻译质量已经接近甚至超过人工翻译
{ {
\footnotesize \footnotesize
\begin{center} \begin{center}
...@@ -273,7 +273,6 @@ ...@@ -273,7 +273,6 @@
\specialrule{0.6pt}{1pt}{1pt} \specialrule{0.6pt}{1pt}{1pt}
人工翻译 & 68.6 & REFERENCE-HT \\ 人工翻译 & 68.6 & REFERENCE-HT \\
& 67.6 & REFERENCE-PE \\ & 67.6 & REFERENCE-PE \\
& 62.1 & REFERENCE-WMT \\
\specialrule{1pt}{1pt}{1pt} \specialrule{1pt}{1pt}{1pt}
\end{tabular}\\ \end{tabular}\\
\addlinespace[-0.3ex] \addlinespace[-0.3ex]
...@@ -685,7 +684,7 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\ ...@@ -685,7 +684,7 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\
\begin{itemize} \begin{itemize}
\item 2015年前统计机器翻译(SMT)在NLP是具有统治力的 \item 2015年前统计机器翻译(SMT)在NLP是具有统治力的
\begin{itemize} \begin{itemize}
\item 当时NMT的系统还很初级，被SMT碾压 \item 当时的NMT系统还很初级，被SMT碾压
\item 大多数的认知还没有进化到NMT时代，甚至Kalchbrenner等人早期的报告也被人质疑 \item 大多数的认知还没有进化到NMT时代，甚至Kalchbrenner等人早期的报告也被人质疑
\end{itemize} \end{itemize}
\item 2016年情况大有改变，当时非常受关注的一项工作是Google上线了神经机器翻译系统GNMT \item 2016年情况大有改变，当时非常受关注的一项工作是Google上线了神经机器翻译系统GNMT
...@@ -708,7 +707,8 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\ ...@@ -708,7 +707,8 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\
\item 解码端也是一个RNN，利用编码结果逐词解码出译文 \item 解码端也是一个RNN，利用编码结果逐词解码出译文
\end{itemize} \end{itemize}
\end{itemize} \end{itemize}
%%% 图
\vspace{-0.5em}
\begin{center} \begin{center}
\begin{tikzpicture} \begin{tikzpicture}
\newlength{\base} \newlength{\base}
...@@ -721,29 +721,31 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\ ...@@ -721,29 +721,31 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\
\begin{scope}[local bounding box=RNNMT] \begin{scope}[local bounding box=RNNMT]
% RNN Encoder % RNN Encoder
\coordinate (eemb0) at (0,0); \coordinate (eemb0) at (0,0);
\foreach \x [count=\y from 0] in {1,2,...,8} \foreach \x [count=\y from 0] in {1,2,...,10}
\node[rnnnode,minimum height=0.1\base,fill=green!30!white,anchor=west] (eemb\x) at ([xshift=0.4\base]eemb\y.east) {}; \node[rnnnode,minimum height=0.5\base,fill=green!30!white,anchor=west] (eemb\x) at ([xshift=0.4\base]eemb\y.east) {};
\foreach \x in {1,2,...,8} \foreach \x in {1,2,...,10}
\node[rnnnode,fill=blue!30!white,anchor=south] (enc\x) at ([yshift=0.5\base]eemb\x.north) {}; \node[rnnnode,fill=blue!30!white,anchor=south] (enc\x) at ([yshift=0.5\base]eemb\x.north) {};
\node[wordnode,left=0.4\base of enc1] (init) {$0$}; \node[wordnode,left=0.4\base of enc1] (init) {$0$};
\node[wordnode,below=0pt of eemb1] () {我们}; \node[wordnode,below=0pt of eemb1] () {你};
\node[wordnode,below=0pt of eemb2] () {感}; \node[wordnode,below=0pt of eemb2] () {知道};
\node[wordnode,below=0pt of eemb3] () {兴趣}; \node[wordnode,below=0pt of eemb3] () {去};
\node[wordnode,below=0pt of eemb4] () {的}; \node[wordnode,below=0pt of eemb4] () {北京站};
\node[wordnode,below=0pt of eemb5] () {是}; \node[wordnode,below=0pt of eemb5] () {的};
\node[wordnode,below=0pt of eemb6] () {气候}; \node[wordnode,below=0pt of eemb6] () {路};
\node[wordnode,below=0pt of eemb7] () {。}; \node[wordnode,below=0pt of eemb7] () {怎么};
\node[wordnode,below=0pt of eemb8] () {$\langle$eos$\rangle$}; \node[wordnode,below=0pt of eemb8] () {走};
\node[wordnode,below=0pt of eemb9] () {吗};
\node[wordnode,below=0pt of eemb10] () {$\langle$eos$\rangle$};
% RNN Decoder % RNN Decoder
\foreach \x in {1,2,...,8} \foreach \x in {1,2,...,10}
\node[rnnnode,minimum height=0.1\base,fill=green!30!white,anchor=south] (demb\x) at ([yshift=2\base]enc\x.north) {}; \node[rnnnode,minimum height=0.5\base,fill=green!30!white,anchor=south] (demb\x) at ([yshift=2\base]enc\x.north) {};
\foreach \x in {1,2,...,8} \foreach \x in {1,2,...,10}
\node[rnnnode,fill=blue!30!white,anchor=south] (dec\x) at ([yshift=0.5\base]demb\x.north) {}; \node[rnnnode,fill=blue!30!white,anchor=south] (dec\x) at ([yshift=0.5\base]demb\x.north) {};
\foreach \x in {1,2,...,8} \foreach \x in {1,2,...,10}
\node[rnnnode,minimum height=0.1\base,fill=red!30!white,anchor=south] (softmax\x) at ([yshift=0.5\base]dec\x.north) {}; \node[rnnnode,minimum height=0.5\base,fill=red!30!white,anchor=south] (softmax\x) at ([yshift=0.5\base]dec\x.north) {};
% Decoder input words % Decoder input words
\node[wordnode,below=0pt of demb1] (decwordin) {$\langle$sos$\rangle$}; \node[wordnode,below=0pt of demb1] (decwordin) {$\langle$sos$\rangle$};
\ExtractX{$(demb2.south)$} \ExtractX{$(demb2.south)$}
...@@ -766,8 +768,14 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\ ...@@ -766,8 +768,14 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\
\node[wordnode,anchor=base] () at (\XCoord,\YCoord) {climate}; \node[wordnode,anchor=base] () at (\XCoord,\YCoord) {climate};
\ExtractX{$(demb8.south)$} \ExtractX{$(demb8.south)$}
\ExtractY{$(decwordin.base)$} \ExtractY{$(decwordin.base)$}
\node[wordnode,anchor=base] () at (\XCoord,\YCoord) {.}; \node[wordnode,anchor=base] () at (\XCoord,\YCoord) {Beijing};
\ExtractX{$(demb9.south)$}
\ExtractY{$(decwordin.base)$}
\node[wordnode,anchor=base] () at (\XCoord,\YCoord) {Railway};
\ExtractX{$(demb10.south)$}
\ExtractY{$(decwordin.base)$}
\node[wordnode,anchor=base] () at (\XCoord,\YCoord) {Station};
% Decoder output words % Decoder output words
\node[wordnode,above=0pt of softmax1] (decwordout) {We}; \node[wordnode,above=0pt of softmax1] (decwordout) {We};
\ExtractX{$(softmax2.north)$} \ExtractX{$(softmax2.north)$}
...@@ -791,7 +799,7 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\ ...@@ -791,7 +799,7 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\
\ExtractX{$(softmax8.north)$} \ExtractX{$(softmax8.north)$}
\ExtractY{$(decwordout.base)$} \ExtractY{$(decwordout.base)$}
\node[wordnode,anchor=base] () at (\XCoord,\YCoord) {$\langle$eos$\rangle$}; \node[wordnode,anchor=base] () at (\XCoord,\YCoord) {$\langle$eos$\rangle$};
% Connections % Connections
\draw[-latex'] (init.east) to (enc1.west); \draw[-latex'] (init.east) to (enc1.west);
\foreach \x in {1,2,...,8} \foreach \x in {1,2,...,8}
...@@ -810,10 +818,16 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\ ...@@ -810,10 +818,16 @@ NLP问题的隐含结构假设 & 无隐含结构假设，端到端学习 \\
\end{scope} \end{scope}
% legend % legend
\begin{scope}[shift={(8\base,0)}] \begin{scope}[shift={(10\base,2.5\base)}]
\node[rnnnode,minimum height=0.1\base,fill=green!30!white,label={[label distance=3pt,font=\scriptsize]0:词嵌入层}] (emb) at (0,0) {}; \node[rnnnode,minimum height=0.5\base,fill=green!30!white,label={[label distance=3pt,font=\scriptsize]0:词嵌入层}] (emb) at (0,0) {};
\node[rnnnode,fill=blue!30!white,anchor=north west,label={[label distance=3pt,font=\scriptsize]0:循环单元}] (rnn) at ([yshift=2.7\base]emb.south west) {}; \node[rnnnode,fill=blue!30!white,anchor=north west,label={[label distance=3pt,font=\scriptsize]0:循环单元}] (rnn) at ([yshift=2\base]emb.south west) {};
\node[rnnnode,minimum height=0.1\base,fill=red!30!white,anchor=north west,label={[label distance=3pt,font=\scriptsize]0:输出层}] (softmax) at ([yshift=2.6\base]rnn.south west) {}; \node[rnnnode,minimum height=0.5\base,fill=red!30!white,anchor=north west,label={[label distance=3pt,font=\scriptsize]0:输出层}] (softmax) at ([yshift=2\base]rnn.south west) {};
\node [anchor=north west] (softmax2) at ([xshift=0.6\base]softmax.south west) {\scriptsize{Softmax}};
\node [anchor=north west] (rnn2) at ([xshift=0.6\base]rnn.south west) {\scriptsize{LSTM}};
\node [anchor=west] (reprlabel) at ([xshift=1em]enc10.east) {\scriptsize{句子表示}};
\draw [->,dashed] (reprlabel.west) -- ([xshift=0.1em]enc10.east);
\node [rnnnode,fill=purple!30!white] at (enc10) {};
\end{scope} \end{scope}
\end{tikzpicture} \end{tikzpicture}
\end{center} \end{center}
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论