Skip to content
项目
群组
代码片段
帮助
当前项目
正在载入...
登录 / 注册
切换导航面板
M
mtbookv2
概览
Overview
Details
Activity
Cycle Analytics
版本库
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Charts
问题
0
Issues
0
列表
Board
标记
里程碑
合并请求
0
Merge Requests
0
CI / CD
CI / CD
流水线
作业
日程表
图表
维基
Wiki
代码片段
Snippets
成员
Collapse sidebar
Close sidebar
活动
图像
聊天
创建新问题
作业
提交
Issue Boards
Open sidebar
NiuTrans
mtbookv2
Commits
30d2c4db
Commit
30d2c4db
authored
Jan 12, 2021
by
孟霞
Browse files
Options
Browse Files
Download
Plain Diff
合并分支 'master' 到 'mengxia'
Master 查看合并请求
!874
parents
8724dcde
d3eef4d5
全部展开
隐藏空白字符变更
内嵌
并排
正在显示
46 个修改的文件
包含
198 行增加
和
178 行删除
+198
-178
Chapter1/Figures/figure-comparison-between-interlingua-based-and-transfer-based-translation.tex
+8
-8
Chapter1/Figures/figure-example-nmt.tex
+3
-3
Chapter1/Figures/figure-example-of-source-structure.tex
+1
-1
Chapter1/Figures/figure-example-rbmt.tex
+1
-1
Chapter1/Figures/figure-example-smt.tex
+4
-4
Chapter1/Figures/figure-four-levels-of-rbmt.tex
+5
-5
Chapter1/Figures/figure-process-of-rule-based-translation.tex
+6
-6
Chapter1/Figures/figure-results-zh-to-en-news-field-translation.tex
+8
-8
Chapter1/Figures/figure-zh-sentences-into-en-sentences.tex
+2
-2
Chapter1/chapter1.tex
+3
-3
Chapter10/chapter10.tex
+2
-2
Chapter11/chapter11.tex
+2
-2
Chapter12/chapter12.tex
+2
-2
Chapter13/Figures/figure-a-predefined-course-planning.tex
+6
-6
Chapter13/Figures/figure-bpe.tex
+19
-18
Chapter13/Figures/figure-computation-of-dropout.tex
+8
-8
Chapter13/Figures/figure-curriculum-learning-framework.tex
+2
-2
Chapter13/Figures/figure-ensemble-knowledge-distillation.tex
+21
-20
Chapter13/Figures/figure-exposure-bias.tex
+6
-6
Chapter13/Figures/figure-framework-of-Adversarial-Neural-machine-translation.tex
+2
-2
Chapter13/Figures/figure-label-smoothing.tex
+6
-6
Chapter13/Figures/figure-network-with-dropout.tex
+1
-1
Chapter13/Figures/figure-of-scheduling-sampling-method.tex
+7
-7
Chapter13/Figures/figure-reinforcement-learning-method-based-on-actor-critic.tex
+2
-2
Chapter13/Figures/figure-underfitting-vs-overfitting.tex
+9
-1
Chapter13/Figures/figure-unk-of-bpe.tex
+1
-1
Chapter13/chapter13.tex
+0
-0
Chapter14/chapter14.tex
+2
-2
Chapter15/chapter15.tex
+2
-2
Chapter16/chapter16.tex
+2
-2
Chapter17/chapter17.tex
+2
-2
Chapter18/Figures/figure-comparison-of-incremental-model-optimization-methods.tex
+13
-3
Chapter18/chapter18.tex
+0
-0
Chapter2/chapter2.tex
+2
-2
Chapter3/chapter3.tex
+2
-2
Chapter4/chapter4.tex
+2
-2
Chapter5/chapter5.tex
+2
-2
Chapter6/chapter6.tex
+2
-2
Chapter7/chapter7.tex
+2
-2
Chapter8/chapter8.tex
+2
-2
Chapter9/chapter9.tex
+2
-2
ChapterAcknowledgement/acknowledgement.tex
+2
-2
ChapterAppend/chapterappend.tex
+13
-13
ChapterPostscript/postscript.tex
+0
-0
ChapterPreface/chapterpreface.tex
+2
-2
mt-book-xelatex.tex
+7
-7
没有找到文件。
Chapter1/Figures/figure-comparison-between-interlingua-based-and-transfer-based-translation.tex
查看文件 @
30d2c4db
...
...
@@ -4,15 +4,15 @@
\begin{tikzpicture}
\begin{scope}
\tikzstyle
{
datanode
}
= [minimum width=5em,minimum height=1.7em,fill=red!20,rounded corners=0.3em];
\tikzstyle
{
modelnode
}
= [minimum width=5em,minimum height=1.7em,fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
decodingnode
}
= [minimum width=5em,minimum height=1.7em,fill=green!20,rounded corners=0.3em];
\tikzstyle
{
datanode
}
= [minimum width=5em,minimum height=1.7em,
draw,thick,
fill=red!20,rounded corners=0.3em];
\tikzstyle
{
modelnode
}
= [minimum width=5em,minimum height=1.7em,
draw,thick,
fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
decodingnode
}
= [minimum width=5em,minimum height=1.7em,
draw,thick,
fill=green!20,rounded corners=0.3em];
\node
[datanode,anchor=north west] (s1) at (0,0)
{{
\small
{
语言1
}}}
;
\node
[datanode,anchor=north] (s2) at ([yshift=-4.5em]s1.south)
{{
\small
{
语言3
}}}
;
\node
[datanode,anchor=west] (s3) at ([xshift=4.5em]s1.east)
{{
\small
{
语言2
}}}
;
\node
[datanode,anchor=north] (s4) at ([yshift=-4.5em]s3.south)
{{
\small
{
语言4
}}}
;
\node
[circle,
anchor=north west,inner sep=2pt,fill=blue!2
0] (m1) at ([xshift=0.8em,yshift=-0.5em]s1.south east)
{{
\small
{
中间语言
}}}
;
\node
[circle,
draw,thick,anchor=north west,inner sep=2pt,fill=blue!3
0] (m1) at ([xshift=0.8em,yshift=-0.5em]s1.south east)
{{
\small
{
中间语言
}}}
;
\draw
[<->,very thick] (s1.south) -- (m1.170);
\draw
[<->,very thick] (s2.north) -- (m1.190);
...
...
@@ -24,11 +24,11 @@
\end{scope}
\begin{scope}
[xshift=
16
em]
\begin{scope}
[xshift=
21
em]
\tikzstyle
{
datanode
}
= [minimum width=5em,minimum height=1.7em,fill=red!20,rounded corners=0.3em];
\tikzstyle
{
modelnode
}
= [minimum width=5em,minimum height=1.7em,fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
decodingnode
}
= [minimum width=5em,minimum height=1.7em,fill=green!20,rounded corners=0.3em];
\tikzstyle
{
datanode
}
= [minimum width=5em,minimum height=1.7em,
draw,thick,
fill=red!20,rounded corners=0.3em];
\tikzstyle
{
modelnode
}
= [minimum width=5em,minimum height=1.7em,
draw,thick,
fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
decodingnode
}
= [minimum width=5em,minimum height=1.7em,
draw,thick,
fill=green!20,rounded corners=0.3em];
\node
[datanode,anchor=north west] (s1) at (0,0)
{{
\small
{
语言1
}}}
;
\node
[datanode,anchor=north] (s2) at ([yshift=-4.5em]s1.south)
{{
\small
{
语言3
}}}
;
...
...
Chapter1/Figures/figure-example-nmt.tex
查看文件 @
30d2c4db
...
...
@@ -49,7 +49,7 @@
\end{scope}
\begin{scope}
[xshift=1
6
.8em,yshift=6em]
\begin{scope}
[xshift=1
8
.8em,yshift=6em]
{
\tikzstyle
{
numbernode
}
= [fill=blue!30,minimum height=0.8em,minimum width=0.8em,circle,draw,inner sep=1pt]
...
...
@@ -94,14 +94,14 @@
\begin{pgfonlayer}
{
background
}
{
\node
[draw,
thin,
minimum width=11em,align=left,rounded corners=1pt,fill=red!5,drop shadow]
[fit = (outputnode) (word01) (word02) ] (netbox)
{}
;
\node
[draw,minimum width=11em,align=left,rounded corners=1pt,fill=red!5,drop shadow]
[fit = (outputnode) (word01) (word02) ] (netbox)
{}
;
\node
[rectangle,draw,thin,inner sep=3pt,rounded corners=1pt]
[fit = (neuron01) (neuron02) (neuron03) (neuron04)] (layer1)
{}
;
\node
[rectangle,draw,thin,inner sep=3pt,rounded corners=1pt]
[fit = (neuron11) (neuron12) (neuron13) (neuron14)] (layer2)
{}
;
\node
[rectangle,draw,thin,inner sep=3pt,rounded corners=1pt]
[fit = (neuron21) (neuron22) (neuron23) (neuron24) (neuron25)] (layer3)
{}
;
}
\end{pgfonlayer}
\draw
[->,very thick,dotted] (enclabel.east) ..controls + (east:
2em) and +(west:2em).. ([yshift=-
1em]netbox.west);
\draw
[->,very thick,dotted] (enclabel.east) ..controls + (east:
4em) and +(west:4em).. ([yshift=
1em]netbox.west);
\node
[anchor=north] (netlabel) at (netbox.south)
{
\footnotesize
{
编码器网络示例
}}
;
}
...
...
Chapter1/Figures/figure-example-of-source-structure.tex
查看文件 @
30d2c4db
...
...
@@ -2,7 +2,7 @@
%%% 句法树(层次短语)
\begin{tikzpicture}
{
\small
\begin{scope}
[sibling distance=25pt, level distance = 2
0
pt]
\begin{scope}
[sibling distance=25pt, level distance = 2
5
pt]
{
\scriptsize
\Tree
[.
\node
(r)
{
IP
}
;
[.
\node
(n11)
{
NP
}
; [.
\node
(n21)
{
PN
}
; [.
\node
(l1)
{
她
}
;]]]
...
...
Chapter1/Figures/figure-example-rbmt.tex
查看文件 @
30d2c4db
...
...
@@ -24,7 +24,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (rule1) (rule2) (rule3) (rule3part2) (rule4) (rule4part2) (rule5) (rule5part2) (rule6) (rule6part2) (rulebaselabel)]
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (rule1) (rule2) (rule3) (rule3part2) (rule4) (rule4part2) (rule5) (rule5part2) (rule6) (rule6part2) (rulebaselabel)]
{}
;
}
\end{pgfonlayer}
}
...
...
Chapter1/Figures/figure-example-smt.tex
查看文件 @
30d2c4db
...
...
@@ -23,7 +23,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (example1) (example1part2) (example2) (example2part2) (example3) (example3part2) (example4) (bidatalabel)(example6)] (bidata)
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (example1) (example1part2) (example2) (example2part2) (example3) (example3part2) (example4) (bidatalabel)(example6)] (bidata)
{}
;
}
\end{pgfonlayer}
...
...
@@ -44,7 +44,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (entry1) (entry2) (entry3) (entry4) (entry5)(entry6)(monodatalabel)] (monodata)
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (entry1) (entry2) (entry3) (entry4) (entry5)(entry6)(monodatalabel)] (monodata)
{}
;
}
\end{pgfonlayer}
...
...
@@ -68,7 +68,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (phrase1) (phrase2) (phrase3) (phrase4) (phrase4part2) (phrase5) (phrase5part2) (phrase6)(phrase7) (phrasetablelabel)] (phrasetable)
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (phrase1) (phrase2) (phrase3) (phrase4) (phrase4part2) (phrase5) (phrase5part2) (phrase6)(phrase7) (phrasetablelabel)] (phrasetable)
{}
;
}
\end{pgfonlayer}
...
...
@@ -88,7 +88,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (ngram1) (ngram2) (ngram3) (ngram4) (lmlabel) (ngrame)] (langaugemodel)
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (ngram1) (ngram2) (ngram3) (ngram4) (lmlabel) (ngrame)] (langaugemodel)
{}
;
}
\end{pgfonlayer}
...
...
Chapter1/Figures/figure-four-levels-of-rbmt.tex
查看文件 @
30d2c4db
...
...
@@ -4,11 +4,11 @@
\begin{tikzpicture}
\begin{scope}
\tikzstyle
{
node1
}
= [minimum width=7em,minimum height=1.7em,fill=red!20,rounded corners=0.3em];
\tikzstyle
{
node2
}
= [minimum width=7em,minimum height=2.5em,fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
node3
}
= [minimum width=7em,minimum height=2.5em,fill=green!20,rounded corners=0.3em];
\tikzstyle
{
node4
}
= [minimum width=7em,minimum height=1.7em,fill=orange!20,rounded corners=0.3em];
\tikzstyle
{
node5
}
= [minimum width=4.5em,minimum height=1.7em,dashed];
\tikzstyle
{
node1
}
= [minimum width=7em,minimum height=1.7em,
draw,thick,
fill=red!20,rounded corners=0.3em];
\tikzstyle
{
node2
}
= [minimum width=7em,minimum height=2.5em,
draw,thick,
fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
node3
}
= [minimum width=7em,minimum height=2.5em,
draw,thick,
fill=green!20,rounded corners=0.3em];
\tikzstyle
{
node4
}
= [minimum width=7em,minimum height=1.7em,
draw,thick,
fill=orange!20,rounded corners=0.3em];
\tikzstyle
{
node5
}
= [minimum width=4.5em,minimum height=1.7em,
thick,
dashed];
\node
[node1,anchor=south west] (s1) at (0,0)
{{
\small
{
源语言词串
}}}
;
\node
[node2,anchor=south] (s2) at ([xshift=2.5em,yshift=2em]s1.north)
{}
;
...
...
Chapter1/Figures/figure-process-of-rule-based-translation.tex
查看文件 @
30d2c4db
...
...
@@ -4,9 +4,9 @@
\begin{tikzpicture}
\begin{scope}
\tikzstyle
{
datanode
}
= [minimum width=7em,minimum height=1.7em,fill=red!20,rounded corners=0.3em];
\tikzstyle
{
modelnode
}
= [minimum width=7em,minimum height=1.7em,fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
decodingnode
}
= [minimum width=7em,minimum height=1.7em,fill=green!20,rounded corners=0.3em];
\tikzstyle
{
datanode
}
= [minimum width=7em,minimum height=1.7em,
thick,draw,
fill=red!20,rounded corners=0.3em];
\tikzstyle
{
modelnode
}
= [minimum width=7em,minimum height=1.7em,
thick,draw,
fill=blue!20,rounded corners=0.3em];
\tikzstyle
{
decodingnode
}
= [minimum width=7em,minimum height=1.7em,
thick,draw,
fill=green!20,rounded corners=0.3em];
\node
[datanode,anchor=north west,minimum height=1.7em,minimum width=8em] (s1) at (0,0)
{{
\small
{
源文句子
}}}
;
\node
[modelnode,anchor=north,minimum height=1.7em,minimum width=8em] (s2) at ([yshift=-1.5em]s1.south)
{{
\small
{
源语词法分析
}}}
;
...
...
@@ -14,14 +14,14 @@
\node
[modelnode,anchor=north,minimum height=1.7em,minimum width=8em] (s4) at ([yshift=-1.5em]s3.south)
{{
\small
{
源语句法分析
}}}
;
\node
[datanode,anchor=north,minimum height=1.7em,minimum width=8em] (s5) at ([yshift=-1.5em]s4.south)
{{
\small
{
源文结构
}}}
;
\node
[datanode,anchor=west,minimum height=1.7em,minimum width=8em] (t1) at ([xshift=
14
em]s1.east)
{{
\small
{
译文句子
}}}
;
\node
[datanode,anchor=west,minimum height=1.7em,minimum width=8em] (t1) at ([xshift=
20
em]s1.east)
{{
\small
{
译文句子
}}}
;
\node
[modelnode,anchor=north,minimum height=1.7em,minimum width=8em] (t2) at ([yshift=-1.5em]t1.south)
{{
\small
{
目标语词法生成
}}}
;
\node
[datanode,anchor=north,minimum height=1.7em,minimum width=8em] (t3) at ([yshift=-1.5em]t2.south)
{{
\small
{
译文词串
}}}
;
\node
[modelnode,anchor=north,minimum height=1.7em,minimum width=8em] (t4) at ([yshift=-1.5em]t3.south)
{{
\small
{
目标语句法生成
}}}
;
\node
[datanode,anchor=north,minimum height=1.7em,minimum width=8em] (t5) at ([yshift=-1.5em]t4.south)
{{
\small
{
译文结构
}}}
;
\node
[decodingnode,anchor=west,minimum height=1.7em,minimum width=
8em] (st1) at ([xshift=2
.5em,yshift=0.85em]s5.east)
{{
\small
{
源语-目标语词汇转换
}}}
;
\node
[decodingnode,anchor=north,minimum height=1.7em,minimum width=
8em
] (st2) at ([yshift=0.05em]st1.south)
{{
\small
{
源语-目标语结构转换
}}}
;
\node
[decodingnode,anchor=west,minimum height=1.7em,minimum width=
13em,inner sep=3pt] (st1) at ([xshift=3
.5em,yshift=0.85em]s5.east)
{{
\small
{
源语-目标语词汇转换
}}}
;
\node
[decodingnode,anchor=north,minimum height=1.7em,minimum width=
13em,inner sep=3pt
] (st2) at ([yshift=0.05em]st1.south)
{{
\small
{
源语-目标语结构转换
}}}
;
\draw
[->,very thick] (s1.south) -- (s2.north);
\draw
[->,very thick] (s2.south) -- (s3.north);
...
...
Chapter1/Figures/figure-results-zh-to-en-news-field-translation.tex
查看文件 @
30d2c4db
...
...
@@ -8,22 +8,22 @@
{
\begin{tikzpicture}
\node
[anchor=south west, fill=blue!
5
0, minimum width=1.1cm, minimum height=2.3cm] (mt) at (1,0)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
机器
}}}}
;
\node
[anchor=south west, fill=red!
5
0, minimum width=1.1cm, minimum height=2.7cm] (human) at ([xshift=0.5cm]mt.south east)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
人
}}}}
;
\node
[anchor=south west, fill=blue!
7
0, minimum width=1.1cm, minimum height=2.3cm] (mt) at (1,0)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
机器
}}}}
;
\node
[anchor=south west, fill=red!
7
0, minimum width=1.1cm, minimum height=2.7cm] (human) at ([xshift=0.5cm]mt.south east)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
人
}}}}
;
\node
[anchor=south] (mtscore) at (mt.north)
{
3.9
}
;
\node
[anchor=south] (humanscore) at (human.north)
{
4.7
}
;
\draw
[->,thick] ([xshift=-0.5cm]mt.south west) -- ([xshift=0.5cm]human.south east);
\draw
[->,thick] ([xshift=-0.5cm]mt.south west) -- ([xshift=-0.5cm,yshift=3.2cm]mt.south west);
\draw
[->,
very
thick] ([xshift=-0.5cm]mt.south west) -- ([xshift=0.5cm]human.south east);
\draw
[->,
very
thick] ([xshift=-0.5cm]mt.south west) -- ([xshift=-0.5cm,yshift=3.2cm]mt.south west);
\node
[anchor=north west] (x1) at ([xshift=0.0cm]human.south east)
{
\footnotesize
{
评价对象
}}
;
\node
[anchor=north east] (y1) at ([xshift=-0.5cm,yshift=3.2cm]mt.south west)
{
\footnotesize
{
打分
}}
;
\node
[anchor=north] (y2) at ([yshift=-0cm]y1.south)
{
\footnotesize
{
(分)
}}
;
\node
[anchor=south west, fill=blue!
50, minimum width=1.1cm, minimum height=1.5cm] (mt1) at ([xshift=13
.0em,yshift=-3.0em]mt.east)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
机器
}}}}
;
\node
[anchor=south west, fill=red!
5
0, minimum width=1.1cm, minimum height=2.7cm] (human1) at ([xshift=0.5cm]mt1.south east)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
人
}}}}
;
\node
[anchor=south west, fill=blue!
70, minimum width=1.1cm, minimum height=1.5cm] (mt1) at ([xshift=16
.0em,yshift=-3.0em]mt.east)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
机器
}}}}
;
\node
[anchor=south west, fill=red!
7
0, minimum width=1.1cm, minimum height=2.7cm] (human1) at ([xshift=0.5cm]mt1.south east)
{{
\color
{
white
}
{
\small\sffamily\bfseries
{
人
}}}}
;
\node
[anchor=south] (mtscore1) at (mt1.north)
{
47
}
;
\node
[anchor=south] (humanscore1) at (human1.north)
{
100
}
;
\draw
[->,thick] ([xshift=-0.5cm]mt1.south west) -- ([xshift=0.5cm]human1.south east);
\draw
[->,thick] ([xshift=-0.5cm]mt1.south west) -- ([xshift=-0.5cm,yshift=3.2cm]mt1.south west);
\draw
[->,
very
thick] ([xshift=-0.5cm]mt1.south west) -- ([xshift=0.5cm]human1.south east);
\draw
[->,
very
thick] ([xshift=-0.5cm]mt1.south west) -- ([xshift=-0.5cm,yshift=3.2cm]mt1.south west);
\node
[anchor=north west] (x1) at ([xshift=0.0cm]human1.south east)
{
\footnotesize
{
评价对象
}}
;
\node
[anchor=north east] (y1) at ([xshift=-0.5cm,yshift=3.2cm]mt1.south west)
{
\footnotesize
{
打分
}}
;
\node
[anchor=north] (y2) at ([yshift=-0cm]y1.south)
{
\footnotesize
{
(分)
}}
;
...
...
Chapter1/Figures/figure-zh-sentences-into-en-sentences.tex
查看文件 @
30d2c4db
...
...
@@ -21,7 +21,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (example1) (example1part2) (example2) (example2part2) (example3) (examplebaselabel)]
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (example1) (example1part2) (example2) (example2part2) (example3) (examplebaselabel)]
{}
;
}
\end{pgfonlayer}
...
...
@@ -40,7 +40,7 @@
\begin{pgfonlayer}
{
background
}
{
\node
[rectangle,draw=ublue, inner sep=0mm]
[fit = (entry1) (entry2) (entry3) (entry4) (dictionarylabel)]
{}
;
\node
[rectangle,draw=ublue,
thick,
inner sep=0mm]
[fit = (entry1) (entry2) (entry3) (entry4) (dictionarylabel)]
{}
;
}
\end{pgfonlayer}
...
...
Chapter1/chapter1.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
@@ -38,7 +38,7 @@
%----------------------------------------------
\begin{figure}
[htp]
\centering
\includegraphics
[scale=0.2]
{
./Chapter1/Figures/figure-zh
_
en-example.png
}
\includegraphics
[scale=0.2
5
]
{
./Chapter1/Figures/figure-zh
_
en-example.png
}
\caption
{
通过计算机将汉语翻译为英语
}
\label
{
fig:1-1
}
\end{figure}
...
...
Chapter10/chapter10.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter11/chapter11.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter12/chapter12.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter13/Figures/figure-a-predefined-course-planning.tex
查看文件 @
30d2c4db
\begin{tikzpicture}
\tikzstyle
{
node1
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
ugreen!10!
blue!10]
\tikzstyle
{
node2
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
ugreen!20!blue!20
]
\tikzstyle
{
node3
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
ugreen!30!blue!3
0]
\tikzstyle
{
node4
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
ugreen!40!blue!40
]
\tikzstyle
{
node5
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
ugreen!50!blue!5
0]
\tikzstyle
{
node1
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=blue!10]
\tikzstyle
{
node2
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
blue!25
]
\tikzstyle
{
node3
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
blue!4
0]
\tikzstyle
{
node4
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
blue!55
]
\tikzstyle
{
node5
}
=[inner sep=0mm,minimum height=1.5em,minimum width=3em,fill=
blue!7
0]
\begin{scope}
\foreach
\i
/
\j
/
\z
in
...
...
@@ -27,7 +27,7 @@
\node
[anchor=north west]
(y1) at ([xshift=14.5em,yshift=-1.3em]n00.south west)
{
训练时长
}
;
\node
[anchor=west,font=\small]
(t1) at ([xshift=0.5em,yshift=0em]n52.east)
{$
\cdots
$}
;
\node
[anchor=west,node1,minimum width=0.5em]
(c1) at ([xshift=
3em,yshift=-2.5em]t1
.east)
{}
;
\node
[anchor=west,node1,minimum width=0.5em]
(c1) at ([xshift=
5em,yshift=0em]n50
.east)
{}
;
\node
[anchor=south,node2,minimum width=0.5em]
(c2) at ([xshift=0em,yshift=0em]c1.north)
{}
;
\node
[anchor=south,node3,minimum width=0.5em]
(c3) at ([xshift=0em,yshift=0em]c2.north)
{}
;
\node
[anchor=south,node4,minimum width=0.5em]
(c4) at ([xshift=0em,yshift=0em]c3.north)
{}
;
...
...
Chapter13/Figures/figure-bpe.tex
查看文件 @
30d2c4db
\begin{tikzpicture}
\tikzstyle
{
tnode
}
= [rectangle,inner sep=0em,minimum width=8em,minimum height=6.6em,rounded corners=5pt,fill=green!5,drop shadow]
\tikzstyle
{
pnode
}
= [rectangle,inner sep=0em,minimum width=8em,minimum height=6.6em,rounded corners=5pt,fill=yellow!5,drop shadow]
\tikzstyle
{
mnode
}
= [rectangle,inner sep=0em,minimum width=8em,minimum height=6.6em,rounded corners=5pt,fill=red!5,drop shadow]
\tikzstyle
{
tnode
}
= [rectangle,inner sep=0em,minimum width=8em,minimum height=6.6em,rounded corners=5pt,fill=green!
1
5,drop shadow]
\tikzstyle
{
pnode
}
= [rectangle,inner sep=0em,minimum width=8em,minimum height=6.6em,rounded corners=5pt,fill=yellow!
1
5,drop shadow]
\tikzstyle
{
mnode
}
= [rectangle,inner sep=0em,minimum width=8em,minimum height=6.6em,rounded corners=5pt,fill=red!
1
5,drop shadow]
\tikzstyle
{
wnode
}
= [inner sep=0em,minimum height=1.5em]
%第一排
...
...
@@ -19,13 +19,13 @@
\begin{pgfonlayer}
{
background
}
\node
[rectangle,inner sep=0.7em,draw,ugreen!60,dashed,
very
thick,rounded corners=7pt] [fit = (n1) (n4)] (box1)
{}
;
\node
[rectangle,inner sep=0.7em,draw,ugreen!60,dashed,thick,rounded corners=7pt] [fit = (n1) (n4)] (box1)
{}
;
\end{pgfonlayer}
\node
[anchor=west,align=left,font=
\footnotesize
] (nt1) at ([xshift=0.1em,yshift=0em]n2.east)
{
统计词表和
\\
[0.5ex]
词频
}
;
\node
[anchor=west,align=left,font=
\footnotesize
] (nt2) at ([xshift=0.1em,yshift=1em]n3.east)
{
按字符切分
}
;
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]n2.east)--([xshift=0em,yshift=0em]n3.west);
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]n3.east)--([xshift=0em,yshift=0em]n4.west);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]n2.east)--([xshift=0em,yshift=0em]n3.west);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]n3.east)--([xshift=0em,yshift=0em]n4.west);
%第二排
\node
[anchor=north west,wnode,font=
\large
] (n5) at ([xshift=0em,yshift=-2em]n2.south west)
{
BPE学习
}
;
...
...
@@ -34,7 +34,7 @@
\node
[anchor=north west,align=left,font=
\footnotesize
] (n61) at ([xshift=0.2em,yshift=-0.2em]n6.north west)
{{
\small
词表
}
\\
l/o/w
\\
l/o/w/e/r
\\
n/e/w/e/s/t
\\
w/i/d/e/s/t
\\
$
\ldots
$}
;
\node
[anchor=north east,align=center,font=
\footnotesize
] (n62) at ([xshift=-0.2em,yshift=-0.2em]n6.north east)
{{
\small
频率
}
\\
6
\\
2
\\
6
\\
3
\\
$
\ldots
$}
;
\draw
[->,
very thick,ublue] ([xshift=-0em,yshift=-0em]n4.south) .. controls +(south:4em) and +(north:4em) .. ([xshift=-0
em,yshift=-0em]n6.north);
\draw
[->,
thick,ublue] ([xshift=-0em,yshift=-0em]n4.south) .. controls +(south:4em) and +(north:4em) .. ([xshift=1
em,yshift=-0em]n6.north);
\node
[anchor=west,pnode] (n7) at ([xshift=5em,yshift=0em]n6.east)
{}
;
\node
[anchor=north west,align=left,font=
\footnotesize
] (n71) at ([xshift=0.2em,yshift=-0.2em]n7.north west)
{{
\small
二元组
}
\\
(e,s)
\\
(s,t)
\\
(l,o)
\\
(o,w)
\\
$
\ldots
$}
;
...
...
@@ -45,15 +45,15 @@
\node
[anchor=west,align=left,font=
\footnotesize
] (nt3) at ([xshift=0.1em,yshift=0em]n6.east)
{
统计二元组
\\
[0.5ex]
的频次
}
;
\node
[anchor=west,align=left,font=
\footnotesize
] (nt4) at ([xshift=0em,yshift=-0.4em]n7.east)
{
频次最高的
\\
[0.5ex]
(e,s)加入表
\\
中
}
;
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]n6.east)--([xshift=0em,yshift=0em]n7.west);
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]n7.east)--([xshift=0em,yshift=0em]n8.west);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]n6.east)--([xshift=0em,yshift=0em]n7.west);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]n7.east)--([xshift=0em,yshift=0em]n8.west);
%第三排
\node
[anchor=north,tnode] (n9) at ([xshift=0em,yshift=-2.5em]n6.south)
{}
;
\node
[anchor=north west,align=left,font=
\footnotesize
] (n91) at ([xshift=0.2em,yshift=-0.2em]n9.north west)
{{
\small
词表
}
\\
l/o/w
\\
l/o/w/e/r
\\
n/e/w/
{
\red
es
}
/t
\\
w/i/d/
{
\red
es
}
/t
\\
$
\ldots
$}
;
\node
[anchor=north east,align=center,font=
\footnotesize
] (n92) at ([xshift=-0.2em,yshift=-0.2em]n9.north east)
{{
\small
频率
}
\\
6
\\
2
\\
6
\\
3
\\
$
\ldots
$}
;
\draw
[->,
very thick,ublue] ([xshift=-0em,yshift=-0em]n8.south) .. controls +(south:3em) and +(north:3em) .. ([xshift=-0
em,yshift=-0em]n9.north);
\draw
[->,
thick,ublue] ([xshift=-0em,yshift=-0em]n8.south) .. controls +(south:3em) and +(north:3em) .. ([xshift=1
em,yshift=-0em]n9.north);
\node
[anchor=north west,ublue,font=
\footnotesize
,align=left] (l1) at ([xshift=1em,yshift=-0em]n7.south east)
{
在词表中
\\
[0.8ex]
合并(e,s)
}
;
\node
[anchor=west,pnode] (n10) at ([xshift=5em,yshift=0em]n9.east)
{}
;
...
...
@@ -65,17 +65,17 @@
\node
[anchor=west,align=left,font=
\footnotesize
] (nt5) at ([xshift=0.1em,yshift=0em]n9.east)
{
统计二元组
\\
[0.5ex]
的频次
}
;
\node
[anchor=west,align=left,font=
\footnotesize
] (nt6) at ([xshift=0em,yshift=-0.4em]n10.east)
{
频次最高的
\\
[0.5ex]
(es,t)加入表
\\
中
}
;
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]n9.east)--([xshift=0em,yshift=0em]n10.west);
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]n10.east)--([xshift=0em,yshift=0em]n11.west);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]n9.east)--([xshift=0em,yshift=0em]n10.west);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]n10.east)--([xshift=0em,yshift=0em]n11.west);
%第四排
\node
[anchor=north,ublue] (cd) at ([xshift=0em,yshift=-
1.5
em]n10.south)
{$
\cdots
$}
;
\draw
[->,
very
thick,ublue] ([xshift=-0em,yshift=-0em]n11.south) .. controls +(south:2em) and +(north:2em) .. ([xshift=-0em,yshift=-0em]cd.north);
\node
[anchor=north west,ublue,font=
\footnotesize
,align=left] (l2) at ([xshift=1em,yshift=
0.6
em]n10.south east)
{
在词表中
\\
[0.8ex]
合并(es,t)
}
;
\node
[anchor=north,ublue] (cd) at ([xshift=0em,yshift=-
2.0
em]n10.south)
{$
\cdots
$}
;
\draw
[->,thick,ublue] ([xshift=-0em,yshift=-0em]n11.south) .. controls +(south:2em) and +(north:2em) .. ([xshift=-0em,yshift=-0em]cd.north);
\node
[anchor=north west,ublue,font=
\footnotesize
,align=left] (l2) at ([xshift=1em,yshift=
-1.0
em]n10.south east)
{
在词表中
\\
[0.8ex]
合并(es,t)
}
;
\node
[anchor=east,ublue,align=left,font=
\footnotesize
] (l3) at ([xshift=-0.5em,yshift=0em]cd.west)
{
直至达到设定的符号合
\\
并表大小或无法合并
}
;
\begin{pgfonlayer}
{
background
}
\node
[rectangle,inner sep=0.7em,draw,orange!40,dashed,
very
thick,rounded corners=7pt] [fit = (n5) (n8) (l3) (cd)] (box2)
{}
;
\node
[rectangle,inner sep=0.7em,draw,orange!40,dashed,thick,rounded corners=7pt] [fit = (n5) (n8) (l3) (cd)] (box2)
{}
;
\end{pgfonlayer}
%第五排
...
...
@@ -90,9 +90,9 @@
\node
[anchor=south west,wnode,font=
\large
] (n13) at ([xshift=0em,yshift=0.5em]n12.north west)
{
输出符号合并表
}
;
\begin{pgfonlayer}
{
background
}
\node
[rectangle,inner sep=0.7em,draw,red!40,dashed,
very
thick,rounded corners=7pt] [fit = (n12) (n13)] (box3)
{}
;
\node
[rectangle,inner sep=0.7em,draw,red!40,dashed,thick,rounded corners=7pt] [fit = (n12) (n13)] (box3)
{}
;
\end{pgfonlayer}
\draw
[->,
very
thick,ublue] ([xshift=0em,yshift=0em]cd.south)--([xshift=0em,yshift=0em]n12.north);
\draw
[->,thick,ublue] ([xshift=0em,yshift=0em]cd.south)--([xshift=0em,yshift=0em]n12.north);
\end{tikzpicture}
\ No newline at end of file
Chapter13/Figures/figure-computation-of-dropout.tex
查看文件 @
30d2c4db
...
...
@@ -18,10 +18,10 @@
\node
[] (standard) at ([xshift=-1em]neuron
_
z.west)
{
\scriptsize
{$
\mathbf
{
w
}_{
i
}^{
l
}$}}
;
\node
[] (standard) at ([xshift=0.6em,yshift=0.3em]neuron
_
z.east)
{
\scriptsize
{$
f
$}}
;
\draw
[->,line width=0.3mm] (neuron
_
b.east) -- (neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (neuron
_
y3.east) -- (neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (neuron
_
y2.east) -- (neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (neuron
_
y1.east) -- (neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (neuron
_
b.east) -- (neuron
_
z.
130
);
\draw
[->,line width=0.3mm] (neuron
_
y3.east) -- (neuron
_
z.
170
);
\draw
[->,line width=0.3mm] (neuron
_
y2.east) -- (neuron
_
z.
190
);
\draw
[->,line width=0.3mm] (neuron
_
y1.east) -- (neuron
_
z.
230
);
\draw
[->,line width=0.3mm] (neuron
_
z.east) -- (neuron
_
y'.west);
%dropout
...
...
@@ -44,10 +44,10 @@
\node
[] (standard) at ([xshift=-1em]drop
_
neuron
_
z.west)
{
\scriptsize
{$
\mathbf
{
w
}_{
i
}^{
l
}$}}
;
\node
[] (standard) at ([xshift=0.6em,yshift=0.3em]drop
_
neuron
_
z.east)
{
\scriptsize
{$
f
$}}
;
%structure
\draw
[->,line width=0.3mm] (drop
_
neuron
_
b.east) -- (drop
_
neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y3'.east) -- (drop
_
neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y2'.east) -- (drop
_
neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y1'.east) -- (drop
_
neuron
_
z.
west
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
b.east) -- (drop
_
neuron
_
z.
130
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y3'.east) -- (drop
_
neuron
_
z.
170
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y2'.east) -- (drop
_
neuron
_
z.
190
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y1'.east) -- (drop
_
neuron
_
z.
230
);
\draw
[->,line width=0.3mm] (drop
_
neuron
_
z.east) -- (drop
_
neuron
_
y'.west);
%r
\draw
[->,line width=0.3mm] (drop
_
neuron
_
y3.east) -- (drop
_
neuron
_
y3'.west);
...
...
Chapter13/Figures/figure-curriculum-learning-framework.tex
查看文件 @
30d2c4db
\begin{tikzpicture}
\tikzstyle
{
node
}
=[inner sep=0mm,minimum height=3em,minimum width=6em,rounded corners=5pt]
\tikzstyle
{
node
}
=[inner sep=0mm,
draw,thick,
minimum height=3em,minimum width=6em,rounded corners=5pt]
\node
[anchor=west,node,fill=ugreen!15]
(n1) at (0,0)
{
训练集
}
;
\node
[anchor=west,node,fill=yellow!15]
(n2) at ([xshift=
2
em,yshift=0em]n1.east)
{
难度评估器
}
;
\node
[anchor=west,node,fill=yellow!15]
(n2) at ([xshift=
4
em,yshift=0em]n1.east)
{
难度评估器
}
;
\node
[anchor=west,node,fill=red!15]
(n3) at ([xshift=4em,yshift=0em]n2.east)
{
训练调度器
}
;
\node
[anchor=west,node,fill=blue!15]
(n4) at ([xshift=4em,yshift=0em]n3.east)
{
模型训练器
}
;
...
...
Chapter13/Figures/figure-ensemble-knowledge-distillation.tex
查看文件 @
30d2c4db
...
...
@@ -6,30 +6,30 @@
\setlength
{
\YShift
}{
0.8
\base
}
\setlength
{
\XShift
}{
0.8
\base
}
\tikzstyle
{
modelnode
}
= [rectangle,draw,
rounded corners=2pt,inner sep=0
pt,minimum height=4.5em,minimum width=2em,font=
\small
,anchor=north]
\tikzstyle
{
modelnode
}
= [rectangle,draw,
thick,rounded corners=2pt,inner sep=2
pt,minimum height=4.5em,minimum width=2em,font=
\small
,anchor=north]
\coordinate
(stu01) at (0,0);
\coordinate
(stu02) at ([xshift=
3
em]stu01);
\coordinate
(stu03) at ([xshift=
3
em]stu02);
\coordinate
(stu04) at ([xshift=
3
em]stu03);
\coordinate
(stu05) at ([xshift=
3
em]stu04);
\coordinate
(tea01) at ([xshift=
8
em]stu05);
\coordinate
(tea02) at ([xshift=
3
em]tea01);
\coordinate
(stu02) at ([xshift=
4
em]stu01);
\coordinate
(stu03) at ([xshift=
4
em]stu02);
\coordinate
(stu04) at ([xshift=
4
em]stu03);
\coordinate
(stu05) at ([xshift=
4
em]stu04);
\coordinate
(tea01) at ([xshift=
10
em]stu05);
\coordinate
(tea02) at ([xshift=
4
em]tea01);
% iterations
\foreach
\curr
/
\prev
in
{
1/0,2/1,3/2
}
{
% models
\node
[modelnode,fill=yellow!20]
(stu
\curr
1) at ([yshift=-
2
em]stu
\prev
1.south)
{
\rotatebox
{
90
}{
学生模型
$
1
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
2) at ([yshift=-
2
em]stu
\prev
2.south)
{
\rotatebox
{
90
}{
学生模型
$
2
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
3) at ([yshift=-
2
em]stu
\prev
3.south)
{
\rotatebox
{
90
}{
学生模型
$
3
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
4) at ([yshift=-
2
em]stu
\prev
4.south)
{
\rotatebox
{
90
}{
学生模型
$
4
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
5) at ([yshift=-
2
em]stu
\prev
5.south)
{
\rotatebox
{
90
}{
学生模型
$
5
$}}
;
\node
[modelnode]
(tea
\curr
1) at ([yshift=-
2
em]tea
\prev
1.south)
{
\rotatebox
{
90
}{
\color
{
red!60
}
教师模型
$
1
$}}
;
\node
[modelnode]
(tea
\curr
2) at ([yshift=-
2
em]tea
\prev
2.south)
{
\rotatebox
{
90
}{
\color
{
blue!60
}
教师模型
$
2
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
1) at ([yshift=-
3
em]stu
\prev
1.south)
{
\rotatebox
{
90
}{
学生模型
$
1
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
2) at ([yshift=-
3
em]stu
\prev
2.south)
{
\rotatebox
{
90
}{
学生模型
$
2
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
3) at ([yshift=-
3
em]stu
\prev
3.south)
{
\rotatebox
{
90
}{
学生模型
$
3
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
4) at ([yshift=-
3
em]stu
\prev
4.south)
{
\rotatebox
{
90
}{
学生模型
$
4
$}}
;
\node
[modelnode,fill=yellow!20]
(stu
\curr
5) at ([yshift=-
3
em]stu
\prev
5.south)
{
\rotatebox
{
90
}{
学生模型
$
5
$}}
;
\node
[modelnode]
(tea
\curr
1) at ([yshift=-
3
em]tea
\prev
1.south)
{
\rotatebox
{
90
}{
\color
{
red!60
}
教师模型
$
1
$}}
;
\node
[modelnode]
(tea
\curr
2) at ([yshift=-
3
em]tea
\prev
2.south)
{
\rotatebox
{
90
}{
\color
{
blue!60
}
教师模型
$
2
$}}
;
% 集成 labels
\draw
[-
latex'
]
([xshift=2pt]stu
\curr
5.east) to node [auto]
{
\small
集成
}
([xshift=-2pt]tea
\curr
1.west);
\draw
[-
>,very thick
]
([xshift=2pt]stu
\curr
5.east) to node [auto]
{
\small
集成
}
([xshift=-2pt]tea
\curr
1.west);
}
% iteration labels
...
...
@@ -37,8 +37,8 @@
\node
[font=\small,anchor=east,purple!80]
(iterate2) at ([xshift=-1em]stu31.west)
{
\rotatebox
{
90
}{
轮数
$
2
$}}
;
% distillation labels
\node
[font=\small,anchor=south west]
(distill1) at ([yshift=
0.8
em]iterate1.north west)
{
知识蒸馏
}
;
\node
[font=\small,anchor=south west]
(distill2) at ([yshift=
0.8
em]iterate2.north west)
{
知识蒸馏
}
;
\node
[font=\small,anchor=south west]
(distill1) at ([yshift=
1.2
em]iterate1.north west)
{
知识蒸馏
}
;
\node
[font=\small,anchor=south west]
(distill2) at ([yshift=
1.2
em]iterate2.north west)
{
知识蒸馏
}
;
% 学生模型 groups
\begin{pgfonlayer}
{
background
}
...
...
@@ -50,10 +50,10 @@
\end{pgfonlayer}
% distillation
\draw
[-latex',red!60,very thick]
(tea11.south) .. controls +(south:1.5em) and +(north:
2
em) .. (group21.north);
\draw
[-latex',blue!60,very thick]
(tea12.south) .. controls +(south:2em) and +(north:
1.5
em) .. (group22.north);
\draw
[-latex',red!60,very thick]
(tea11.south) .. controls +(south:1.5em) and +(north:
3
em) .. (group21.north);
\draw
[-latex',blue!60,very thick]
(tea12.south) .. controls +(south:2em) and +(north:
2
em) .. (group22.north);
\draw
[-latex',red!60,very thick]
(tea21.south) .. controls +(south:2em) and +(north:2.5em) .. (group32.north);
\draw
[-latex',blue!60,very thick]
(tea22.south) .. controls +(south:2em) and +(north:
1.5
em) .. (group31.north);
\draw
[-latex',blue!60,very thick]
(tea22.south) .. controls +(south:2em) and +(north:
3
em) .. (group31.north);
\end{tikzpicture}
\ No newline at end of file
Chapter13/Figures/figure-exposure-bias.tex
查看文件 @
30d2c4db
...
...
@@ -8,8 +8,8 @@
\begin{scope}
[]
\tikzstyle
{
rnnnode
}
= [draw,inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=red!15]
\tikzstyle
{
snode
}
= [draw,inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=blue!15]
\tikzstyle
{
rnnnode
}
= [draw,
thick,
inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=red!15]
\tikzstyle
{
snode
}
= [draw,
thick,
inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=blue!15]
\tikzstyle
{
ynode
}
= [inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt]
...
...
@@ -122,8 +122,8 @@
\begin{scope}
[yshift=-2in]
\tikzstyle
{
rnnnode
}
= [draw,inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=red!15]
\tikzstyle
{
snode
}
= [draw,inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=blue!15]
\tikzstyle
{
rnnnode
}
= [draw,
thick,
inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=red!15]
\tikzstyle
{
snode
}
= [draw,
thick,
inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt,fill=blue!15]
\tikzstyle
{
ynode
}
= [inner sep=2pt,minimum width=3em,minimum height=1.5em,rounded corners=1pt]
...
...
@@ -166,8 +166,8 @@
\draw
[->,thick,dotted] ([xshift=-2.5em,yshift=1em]n11.north) .. controls +(south:2em) and +(west:0.1em) .. ([xshift=0.2em,yshift=0em]n11.west);
\draw
[->,thick,dotted] ([xshift=0em,yshift=0em]n14.east) .. controls +(east:0.3em) and +(north:
2
em) .. ([xshift=2.5em,yshift=-0.5em]n14.south);
\draw
[->,thick,dotted] ([xshift=0em,yshift=0em]n6.east) .. controls
([xshift=2em,yshift=1em]n6.east) and ([xshift=-2em,yshift=-2em]n4.south west
) .. ([xshift=0.2em,yshift=-0em]n12.west);
\draw
[->,thick,dotted] ([xshift=0em,yshift=0em]n14.east) .. controls +(east:0.3em) and +(north:
1.5
em) .. ([xshift=2.5em,yshift=-0.5em]n14.south);
\draw
[->,thick,dotted] ([xshift=0em,yshift=0em]n6.east) .. controls
+(east:2em) and +(west:2em
) .. ([xshift=0.2em,yshift=-0em]n12.west);
\draw
[->,thick] ([xshift=-0em,yshift=0em]n11.north)--([xshift=-0em,yshift=0em]n3.south);
\draw
[->,thick] ([xshift=0em,yshift=0em]n12.north)--([xshift=-0em,yshift=-0em]n4.south);
...
...
Chapter13/Figures/figure-framework-of-Adversarial-Neural-machine-translation.tex
查看文件 @
30d2c4db
...
...
@@ -9,9 +9,9 @@
\node
[anchor=west,wode] (n1) at (0,0)
{$
y
$}
;
\node
[anchor=north west,wode] (n2) at ([xshift=3em,yshift=-2.5em]n1.south east)
{$
x
$}
;
\node
[anchor=south west,rnnnode] (n3) at ([xshift=8em,yshift=0.5em]n2.north east)
{
生成模型
$
G
$}
;
\node
[anchor=south west,rnnnode
,thick
] (n3) at ([xshift=8em,yshift=0.5em]n2.north east)
{
生成模型
$
G
$}
;
\node
[anchor=south east,wode] (n4) at ([xshift=-2em,yshift=0em]n3.north west)
{$
\hat
{
y
}$}
;
\node
[anchor=south,snode] (n5) at ([xshift=0em,yshift=6em]n2.north)
{
判别网络
$
D
$}
;
\node
[anchor=south,snode
,thick
] (n5) at ([xshift=0em,yshift=6em]n2.north)
{
判别网络
$
D
$}
;
\node
[anchor=west,align=left,font=
\small
] (n6) at ([xshift=15em,yshift=-3em]n5.east)
{
根据
$
(
\seq
{
x
}
,
\seq
{
\hat
{
y
}}
)
$
生
\\
成奖励信号
}
;
...
...
Chapter13/Figures/figure-label-smoothing.tex
查看文件 @
30d2c4db
...
...
@@ -55,18 +55,18 @@
\node
[anchor=south,font=
\scriptsize
] (w8) at (label7.north)
{$
0
.
1
$}
;
\node
[font=\scriptsize]
(line1) at ([xshift=
9
em,yshift=-1.5em]model
_
label7.east)
{$
Loss
=-
0
.
3
\log
p
_{
3
}
-
\sum
_{
i
=
1
}^{
7
}
0
.
1
\log
p
_{
i
}$}
;
\node
[font=\scriptsize]
(line2) at ([xshift=
5.9
em,yshift=3em]model
_
label7.east)
{$
Loss
=-
\log
p
_{
3
}$}
;
\node
[font=\scriptsize]
(line1) at ([xshift=
13
em,yshift=-1.5em]model
_
label7.east)
{$
Loss
=-
0
.
3
\log
p
_{
3
}
-
\sum
_{
i
=
1
}^{
7
}
0
.
1
\log
p
_{
i
}$}
;
\node
[font=\scriptsize]
(line2) at ([xshift=
9.5
em,yshift=3em]model
_
label7.east)
{$
Loss
=-
\log
p
_{
3
}$}
;
\begin{pgfonlayer}
{
background
}
\node
[rectangle,inner sep=0.5em,rounded corners=1pt,very thick,dotted,draw=red] [fit =(model
_
w3) (model
_
label1) (model
_
label7) (one
_
hot
_
w3)] (box1)
{}
;
\node
[rectangle,inner sep=0.2em,rounded corners=1pt,fill=purple!10,drop shadow,draw=purple] [fit = (line2)] (box3)
{}
;
\draw
[->,dotted,very thick,red] ([yshift=
1em]box1.east) .. controls +(east:1) and +(west:1
) .. (box3.west);
\node
[rectangle,
thick,
inner sep=0.2em,rounded corners=1pt,fill=purple!10,drop shadow,draw=purple] [fit = (line2)] (box3)
{}
;
\draw
[->,dotted,very thick,red] ([yshift=
2em]box1.east) .. controls +(east:1.7) and +(west:1.5
) .. (box3.west);
\node
[rectangle,inner sep=0.7em,rounded corners=1pt,very thick,dotted,draw=ugreen] [fit =(w3) (label1) (label7) (model
_
w3)] (box2)
{}
;
\node
[rectangle,inner sep=0.2em,rounded corners=1pt,fill=green!10,drop shadow,draw=ugreen] [fit = (line1)] (box4)
{}
;
\draw
[->,dotted,very thick,ugreen] ([yshift=-
1.5em]box2.east) .. controls +(east:1) and +(west:1
) .. (box4.west);
\node
[rectangle,
thick,
inner sep=0.2em,rounded corners=1pt,fill=green!10,drop shadow,draw=ugreen] [fit = (line1)] (box4)
{}
;
\draw
[->,dotted,very thick,ugreen] ([yshift=-
2.5em]box2.east) .. controls +(east:1.7) and +(west:1.5
) .. (box4.west);
\end{pgfonlayer}
...
...
Chapter13/Figures/figure-network-with-dropout.tex
查看文件 @
30d2c4db
...
...
@@ -2,7 +2,7 @@
\begin{tikzpicture}
\def\neuronsep
{
1
}
\tikzstyle
{
neuronnode
}
= [minimum size=1
.0
em,circle,draw,thick,ublue,inner sep=1pt, fill=white,align=center]
\tikzstyle
{
neuronnode
}
= [minimum size=1em,circle,draw,thick,ublue,inner sep=1pt, fill=white,align=center]
%standard
...
...
Chapter13/Figures/figure-of-scheduling-sampling-method.tex
查看文件 @
30d2c4db
...
...
@@ -4,8 +4,8 @@
\begin{tikzpicture}
\tikzstyle
{
rnnnode
}
= [draw,inner sep=2pt,minimum width=4em,minimum height=2em,rounded corners=1pt,fill=red!15]
\tikzstyle
{
snode
}
= [draw,inner sep=2pt,minimum width=4em,minimum height=2em,rounded corners=1pt,fill=blue!15]
\tikzstyle
{
rnnnode
}
= [draw,
thick,
inner sep=2pt,minimum width=4em,minimum height=2em,rounded corners=1pt,fill=red!15]
\tikzstyle
{
snode
}
= [draw,
thick,
inner sep=2pt,minimum width=4em,minimum height=2em,rounded corners=1pt,fill=blue!15]
\tikzstyle
{
ynode
}
= [inner sep=2pt,minimum width=4em,minimum height=2em,rounded corners=1pt]
...
...
@@ -14,18 +14,18 @@
\node
[anchor=west,rnnnode] (n3) at ([xshift=3em,yshift=0em]n2.east)
{$
\mathbi
{
h
}_{
j
-
1
}$}
;
\node
[anchor=west,rnnnode] (n4) at ([xshift=3em,yshift=0em]n3.east)
{$
\mathbi
{
h
}_{
j
}$}
;
\node
[anchor=south,snode] (n5) at ([xshift=0em,yshift=1em]n3.north)
{
Softmax
}
;
\node
[anchor=south,ynode] (n6) at ([xshift=0em,yshift=1em]n5.north)
{$
\hat
{{
y
}}_{
j
-
1
}$}
;
\node
[anchor=south,ynode
,fill=green!20
] (n6) at ([xshift=0em,yshift=1em]n5.north)
{$
\hat
{{
y
}}_{
j
-
1
}$}
;
\node
[anchor=south,snode] (n7) at ([xshift=0em,yshift=1em]n4.north)
{
Softmax
}
;
\node
[anchor=south,ynode] (n8) at ([xshift=0em,yshift=1em]n7.north)
{$
\hat
{{
y
}}_{
j
}$}
;
\node
[anchor=south,snode
,font=
\footnotesize
] (n13) at ([xshift=0em,yshift=1em]n1.north)
{
Softmax
}
;
\node
[anchor=south,snode] (n13) at ([xshift=0em,yshift=1em]n1.north)
{
Softmax
}
;
\node
[anchor=south,ynode] (n14) at ([xshift=0em,yshift=1em]n13.north)
{$
\hat
{{
y
}}_{
1
}$}
;
\node
[anchor=north] (x1) at ([xshift=0em,yshift=-1em]n1.south)
{$
\langle
$
sos
$
\rangle
$}
;
\node
[anchor=north,font=
\small
] (x2) at ([xshift=-1.3em,yshift=-2.3em]n3.south)
{$
\hat
{{
y
}}_{
j
-
2
}$}
;
\node
[anchor=north,font=
\small
] (x3) at ([xshift=1.3em,yshift=-2.5em]n3.south)
{${
y
}_{
j
-
2
}$}
;
\node
[anchor=north,font=
\small
] (x4) at ([xshift=1.3em,yshift=-2.5em]n4.south)
{${
y
}_{
j
-
1
}$}
;
\node
[anchor=north,font=
\small
] (x5) at ([xshift=-1.3em,yshift=-2.3em]n4.south)
{$
\hat
{{
y
}}_{
j
-
1
}$}
;
\node
[anchor=north,font=
\small
,fill=green!20
] (x5) at ([xshift=-1.3em,yshift=-2.3em]n4.south)
{$
\hat
{{
y
}}_{
j
-
1
}$}
;
\node
[anchor=south,inner sep=2pt] (st1) at (n6.north)
{
\scriptsize
{
\textbf
{
[step
$
j
-
1
$
]
}}}
;
\node
[anchor=south,inner sep=2pt] (st2) at (n8.north)
{
\scriptsize
{
\textbf
{
[step
$
j
$
]
}}}
;
...
...
@@ -74,7 +74,7 @@
\draw
[-,thick] ([xshift=0em,yshift=0.2em]x4.north)-- ([xshift=1.3em,yshift=0em]n4.south);
\draw
[->,thick,dotted] ([xshift=-2.5em,yshift=1em]x2.north) .. controls +(south:2em) and +(west:0.1em) .. ([xshift=0.2em,yshift=0em]x2.west);
\draw
[->,thick,dotted
] ([xshift=0em,yshift=0em]n6.east) .. controls ([xshift=2em,yshift=1em]n6.east) and ([xshift=-2.5em,yshift=-4em]n4.south west
) .. ([xshift=-0em,yshift=-0em]x5.west);
\draw
[->,thick,dotted] ([xshift=0em,yshift=0em]n14.east) .. controls +(east:0.3em) and +(north:
2
em) .. ([xshift=3em,yshift=-0.5em]n14.south);
\draw
[->,thick,dotted
,ugreen] ([xshift=0em,yshift=0em]n6.east) .. controls +(east:2.5em) and +(west:2.5em
) .. ([xshift=-0em,yshift=-0em]x5.west);
\draw
[->,thick,dotted] ([xshift=0em,yshift=0em]n14.east) .. controls +(east:0.3em) and +(north:
1.5
em) .. ([xshift=3em,yshift=-0.5em]n14.south);
\end{tikzpicture}
Chapter13/Figures/figure-reinforcement-learning-method-based-on-actor-critic.tex
查看文件 @
30d2c4db
\begin{tikzpicture}
\node
[anchor=west,inner sep=0mm,minimum height=4em,minimum width=5.5em,rounded corners=15pt,align=left,draw,fill=red!15]
(n1) at (0,0)
{
Decoder
\\
Encoder
}
;
\node
[anchor=west,
thick,
inner sep=0mm,minimum height=4em,minimum width=5.5em,rounded corners=15pt,align=left,draw,fill=red!15]
(n1) at (0,0)
{
Decoder
\\
Encoder
}
;
\node
[anchor=west,inner sep=0mm,minimum height=4em,minimum width=5.5em,rounded corners=15pt,align=left,draw,fill=green!15]
(n2) at ([xshift=10em,yshift=0em]n1.east)
{
Decoder
\\
Encoder
}
;
\node
[anchor=west,
thick,
inner sep=0mm,minimum height=4em,minimum width=5.5em,rounded corners=15pt,align=left,draw,fill=green!15]
(n2) at ([xshift=10em,yshift=0em]n1.east)
{
Decoder
\\
Encoder
}
;
\node
[anchor=south,inner sep=0mm,font=\small]
(a1) at ([xshift=0em,yshift=1em]n1.north)
{
演员
$
p
$}
;
...
...
Chapter13/Figures/figure-underfitting-vs-overfitting.tex
查看文件 @
30d2c4db
\begin{tabular}
{
l l l
}
\begin{tabular}
{
l l l
l l
}
\begin{tikzpicture}
\draw
[->, thick]
(0,0) to (3,0);
\draw
[->, thick]
(0,-0) to (0,2);
...
...
@@ -25,6 +25,10 @@
\node
[font=
\footnotesize
] at (1.5,-0.5)
{
欠拟合
}
;
\end{tikzpicture}
&
\hspace
{
3em
}
&
\begin{tikzpicture}
\draw
[->, thick]
(0,0) to (3,0);
\draw
[->, thick]
(0,-0) to (0,2);
...
...
@@ -52,6 +56,10 @@
\node
[font=
\footnotesize
] at (1.5,-0.5)
{
拟合合适
}
;
\end{tikzpicture}
&
\hspace
{
3em
}
&
\begin{tikzpicture}
\draw
[->, thick]
(0,0) to (3,0);
\draw
[->, thick]
(0,-0) to (0,2);
...
...
Chapter13/Figures/figure-unk-of-bpe.tex
查看文件 @
30d2c4db
\begin{tikzpicture}
\node
[rounded corners=3pt,minimum width=1.0em,minimum height=2.0em,font=\scriptsize,fill=
green!5
,drop shadow,thick]
(top) at (0,0)
{
\node
[rounded corners=3pt,minimum width=1.0em,minimum height=2.0em,font=\scriptsize,fill=
red!10
,drop shadow,thick]
(top) at (0,0)
{
\begin{tabular}
{
lllllll
}
\multicolumn
{
7
}{
c
}{
符号合并表
}
\\
r
\
$
<
$
e
$
>
$
,
&
e
\
s,
&
l
\
o,
&
es
\
t,
&
lo
\
w,
&
est
\
$
<
$
e
$
>
$
,
&
e
\
r
$
<
$
e
$
>
$
,
...
...
Chapter13/chapter13.tex
查看文件 @
30d2c4db
This source diff could not be displayed because it is too large. You can
view the blob
instead.
Chapter14/chapter14.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter15/chapter15.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter16/chapter16.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter17/chapter17.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter18/Figures/figure-comparison-of-incremental-model-optimization-methods.tex
查看文件 @
30d2c4db
...
...
@@ -7,19 +7,29 @@
% ,minimum height =1em,minimum width=2em
\tikzstyle
{
model
}
= [draw,black,very thick,inner sep=3.5pt,rounded corners=4pt,fill=blue!20,minimum width=4em,minimum height=1.5em,font=
\footnotesize
]
\tikzstyle
{
data
}
= [draw,black,very thick,inner sep=3.5pt,rounded corners=4pt,fill=green!20,minimum width=4em,minimum height=1.5em,font=
\footnotesize
]
\tikzstyle
{
data2
}
= [draw,black,very thick,inner sep=3.5pt,rounded corners=4pt,minimum width=2em,minimum height=1.5em,font=
\footnotesize
]
\tikzstyle
{
word
}
= [inner sep=3.5pt,font=
\footnotesize
]
\node
[data]
(old) at (0,0)
{
旧数据
}
;
\node
[data]
(new) at ([xshift=3em]old.east)
{
新数据
}
;
\node
[data2,fill=green!60]
(old) at (0,0)
{}
;
\node
[data2,fill=green!20]
(old2) at ([xshift=0.9em]old.east)
{}
;
\node
[word]
at (0.1,0.5)
{
旧数据
}
;
\node
[data]
(new) at ([xshift=3em]old2.east)
{
新数据
}
;
\node
[data]
(all) at ([xshift=2.55em,yshift=-4em]old.south)
{
最终数据
}
;
\node
[model]
(final
_
model) at ([xshift=0em,yshift=-4em]all.south)
{
最终模型
}
;
\draw
[->,thick] ([yshift=-0.2em]old.south) .. controls +(south:2.5em) and +(north:2.5em) .. ([xshift=-0.2em,yshift=0.2em]all.north);
\draw
[->,thick] ([yshift=-0.2em]old
2
.south) .. controls +(south:2.5em) and +(north:2.5em) .. ([xshift=-0.2em,yshift=0.2em]all.north);
\draw
[->,thick] ([yshift=-0.2em]new.south) .. controls +(south:2.5em) and +(north:2.5em) .. ([xshift=0.2em,yshift=0.2em]all.north);
\draw
[->,thick] ([yshift=-0.2em]all.south) -- ([yshift=0.2em]final
_
model.north)node[pos=0.5,right,align=center,font=
\footnotesize
]
{
训练
}
;
\node
[word]
at ([yshift=-2em]final
_
model.south)
{
(a)数据混合
}
;
\begin{pgfonlayer}
{
background
}
\node
[rectangle,inner sep=0.5em,rounded corners=1pt,minimum width=4.5em,minimum height=3.2em,draw=red!50,dotted,very thick] at (0.37,0.15)
{}
;
\end{pgfonlayer}
\end{scope}
\end{tikzpicture}
&
...
...
Chapter18/chapter18.tex
查看文件 @
30d2c4db
差异被折叠。
点击展开。
Chapter2/chapter2.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter3/chapter3.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter4/chapter4.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter5/chapter5.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter6/chapter6.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter7/chapter7.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter8/chapter8.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
Chapter9/chapter9.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
ChapterAcknowledgement/acknowledgement.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
ChapterAppend/chapterappend.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
@@ -44,7 +44,7 @@
\vspace
{
0.5em
}
\item
Joshua。Joshua
\upcite
{
Li2010Joshua
}
是由约翰霍普金斯大学的语言和语音处理中心开发的层次短语翻译系统。由于Joshua是由Java语言开发,所以它在不同的平台上运行或开发时具有良好的可扩展性和可移植性。Joshua也是使用非常广泛的开源机器翻译系统之一。网址:
\url
{
https://cwiki.apache.org/confluence/display/JOSHUA/
}
\vspace
{
0.5em
}
\item
SilkRoad。SilkRoad是由五个国内机构(中科院计算所、中科院软件所、中科院自动化所、厦门大学和哈尔滨工业大学)联合开发的基于短语的统计机器翻译系统。该系统是中国乃至亚洲地区第一个开源的统计机器翻译系统。SilkRoad支持多种解码器和规则提取模块,这样可以组合成不同的系统,提供多样的选择。
网址:
\url
{
http://www.nlp.org.cn/project/project.php?projid=14
}
\item
SilkRoad。SilkRoad是由五个国内机构(中科院计算所、中科院软件所、中科院自动化所、厦门大学和哈尔滨工业大学)联合开发的基于短语的统计机器翻译系统。该系统是中国乃至亚洲地区第一个开源的统计机器翻译系统。SilkRoad支持多种解码器和规则提取模块,这样可以组合成不同的系统,提供多样的选择。
\vspace
{
0.5em
}
\item
SAMT。SAMT
\upcite
{
zollmann2007the
}
是由卡内基梅隆大学机器翻译团队开发的语法增强的统计机器翻译系统。SAMT在解码的时候使用目标树来生成翻译规则,而不严格遵守目标语言的语法。SAMT 的一个亮点是它提供了简单但高效的方式在机器翻译中使用句法信息。由于SAMT在hadoop中实现,它可受益于大数据集的分布式处理。网址:
\url
{
http://www.cs.cmu.edu/~zollmann/samt/
}
\vspace
{
0.5em
}
...
...
@@ -151,11 +151,11 @@
\begin{tabular}
{
p
{
1.6cm
}
| p
{
1.2cm
}
p
{
1.6cm
}
p
{
2.6cm
}
p
{
3.9cm
}}
{
任务
}
&
{
语种
}
&{
领域
}
&{
描述
}
&{
数据集地址
}
\\
\hline
\rule
{
0pt
}{
15pt
}
WMT
&
En
Zh
&
新闻、医学
&
以英语为核心的多
&
{
http://www.statmt.org/wmt19/
}
\\
&
De Ru
等
&
、翻译
&
语种机器翻译数据
&
\\
\rule
{
0pt
}{
15pt
}
WMT
&
En
-
Zh
&
新闻、医学
&
以英语为核心的多
&
{
http://www.statmt.org/wmt19/
}
\\
&
En-De
等
&
、翻译
&
语种机器翻译数据
&
\\
&
&
&
集,涉及多种任务
&
\\
\rule
{
0pt
}{
15pt
}
IWSLT
&
En
De Fr
&
口语翻译
&
文本翻译数据集来
&
{
https://wit3.fbk.eu/
}
\\
&
Cs
Zh等
&
&
自TED演讲,数
&
\\
\rule
{
0pt
}{
15pt
}
IWSLT
&
En
-De
&
口语翻译
&
文本翻译数据集来
&
{
https://wit3.fbk.eu/
}
\\
&
En-
Zh等
&
&
自TED演讲,数
&
\\
&
&
&
据规模较小
&
\\
\rule
{
0pt
}{
15pt
}
NIST
&
Zh-En等
&
新闻翻译
&
评测集包括4句参
&
{
https://www.ldc.upenn.edu/coll
}
\\
&
Cs Zh等
&
&
考译文,质量较高
&
aborations/evaluations/nist
\\
...
...
@@ -300,19 +300,19 @@ c(s_u|t_v,\seq{s},\seq{t}) &\approx & \sum_{\seq{a} \in S}\big[\funp{P}_{\theta}
\parinterval
可以以同样的方式修改公式
\eqref
{
eq:1.3
}
-
\eqref
{
eq:1.6
}
的修改结果。进一步,在IBM模型3中,可以定义
$
S
$
如下:
\begin{eqnarray}
S
&
=
&
N(b
^{
\infty
}
(V(
\seq
{
s
}
|
\seq
{
t
}
;2)))
\cup
(
\mathop
{
\cup
}
\limits
_{
ij
}
N(b
_{
i
\leftrightarrow
j
}^{
\infty
}
(V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
,
2))))
S
&
=
&
N(b
^{
\infty
}
(V(
\seq
{
s
}
|
\seq
{
t
}
;2)))
\cup
(
\mathop
{
\cup
}
\limits
_{
ij
}
N(b
_{
i
\leftrightarrow
j
}^{
\infty
}
(V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
;
2))))
\label
{
eq:1.12
}
\end{eqnarray}
\parinterval
为了理解这个公式,先介绍几个概念。
\begin{itemize}
\item
$
V
(
\seq
{
s
}
|
\seq
{
t
}
)
$
表示Viterbi词对齐,
$
V
(
\seq
{
s
}
|
\seq
{
t
}
,
1
)
$
、
$
V
(
\seq
{
s
}
|
\seq
{
t
}
,
2
)
$
和
$
V
(
\seq
{
s
}
|
\seq
{
t
}
,
3
)
$
就分别对应了模型1、2 和3 的Viterbi 词对齐;
\item
把那些满足第
$
j
$
个源语言单词对应第
$
i
$
个目标语言单词(
$
a
_
j
=
i
$
)的词对齐构成的集合记为
$
\seq
{
a
}_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
,
\seq
{
t
}
)
$
。通常称这些对齐中
$
j
$
和
$
i
$
被``钉''在了一起。在
$
\seq
{
a
}_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
,
\seq
{
t
}
)
$
中使
$
\funp
{
P
}
(
\seq
{
a
}
|
\seq
{
s
}
,
\seq
{
t
}
)
$
达到最大的那个词对齐被记为
$
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
)
$
;
\item
$
V
(
\seq
{
s
}
|
\seq
{
t
}
)
$
表示Viterbi词对齐,
$
V
(
\seq
{
s
}
|
\seq
{
t
}
;
1
)
$
、
$
V
(
\seq
{
s
}
|
\seq
{
t
}
;
2
)
$
和
$
V
(
\seq
{
s
}
|
\seq
{
t
}
;
3
)
$
就分别对应了模型1、2 和3 的Viterbi 词对齐;
\item
把那些满足第
$
j
$
个源语言单词对应第
$
i
$
个目标语言单词(
$
a
_
j
=
i
$
)的词对齐构成的集合记为
$
\seq
{
a
}_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
,
\seq
{
t
}
)
$
。通常称这些对齐中
$
j
$
和
$
i
$
被``钉''在了一起。在
$
\seq
{
a
}_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
,
\seq
{
t
}
)
$
中使
$
\funp
{
P
}
(
\seq
{
s
}
,
\seq
{
a
}
|
\seq
{
t
}
)
$
达到最大的那个词对齐被记为
$
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
)
$
;
\item
如果两个词对齐,通过交换两个词对齐连接就能互相转化,则称它们为邻居。一个词对齐
$
\seq
{
a
}$
的所有邻居记为
$
N
(
\seq
{
a
}
)
$
。
\end{itemize}
\vspace
{
0.5em
}
\parinterval
公式
\eqref
{
eq:1.12
}
中,
$
b
^{
\infty
}
(
V
(
\seq
{
s
}
|
\seq
{
t
}
;
2
))
$
和
$
b
_{
i
\leftrightarrow
j
}^{
\infty
}
(
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
,
2
))
$
分别是对
$
V
(
\seq
{
s
}
|
\seq
{
t
}
;
3
)
$
和
$
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
,
3
)
$
的
估计。在计算
$
S
$
的过程中,需要知道一个对齐
$
\seq
{
a
}$
的邻居
$
\seq
{
a
}
'
$
的概率,即通过
$
\funp
{
P
}_{
\theta
}
(
\seq
{
a
}
,
\seq
{
s
}
|
\seq
{
t
}
)
$
计算
$
\funp
{
P
}_{
\theta
}
(
\seq
{
a
}
',
\seq
{
s
}
|
\seq
{
t
}
)
$
。在模型3中,如果
$
\seq
{
a
}$
和
$
\seq
{
a
}
'
$
仅区别于某个源语言单词
$
s
_
j
$
对齐从
$
a
_
j
$
变到
$
a
_{
j
}
'
$
,且
$
a
_
j
$
和
$
a'
_
j
$
均不为零,令
$
a
_
j
=
i
$
,
$
a'
_{
j
}
=
i'
$
,那么
\parinterval
公式
\eqref
{
eq:1.12
}
中,
应该使用
$
V
(
\seq
{
s
}
|
\seq
{
t
}
;
3
)
$
和
$
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
;
3
)
$
进行计算,但其复杂度较高,因此使用
$
b
^{
\infty
}
(
V
(
\seq
{
s
}
|
\seq
{
t
}
;
2
))
$
和
$
b
_{
i
\leftrightarrow
j
}^{
\infty
}
(
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
;
2
))
$
分别对
$
V
(
\seq
{
s
}
|
\seq
{
t
}
;
3
)
$
和
$
V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
;
3
)
$
进行
估计。在计算
$
S
$
的过程中,需要知道一个对齐
$
\seq
{
a
}$
的邻居
$
\seq
{
a
}
'
$
的概率,即通过
$
\funp
{
P
}_{
\theta
}
(
\seq
{
a
}
,
\seq
{
s
}
|
\seq
{
t
}
)
$
计算
$
\funp
{
P
}_{
\theta
}
(
\seq
{
a
}
',
\seq
{
s
}
|
\seq
{
t
}
)
$
。在模型3中,如果
$
\seq
{
a
}$
和
$
\seq
{
a
}
'
$
仅区别于某个源语言单词
$
s
_
j
$
对齐从
$
a
_
j
$
变到
$
a
_{
j
}
'
$
,且
$
a
_
j
$
和
$
a'
_
j
$
均不为零,令
$
a
_
j
=
i
$
,
$
a'
_{
j
}
=
i'
$
,那么
\begin{eqnarray}
\funp
{
P
}_{
\theta
}
(
\seq
{
a
}
',
\seq
{
s
}
|
\seq
{
t
}
)
&
=
&
\funp
{
P
}_{
\theta
}
(
\seq
{
a
}
,
\seq
{
s
}
|
\seq
{
t
}
)
\cdot
\nonumber
\\
...
...
@@ -381,7 +381,7 @@ z_{>1}(\Delta_j|cb;\seq{a},\seq{s},\seq{t}) & = & \sum_{i=1}^l \big[\varepsilon(
\parinterval
模型4 需要像模型3 一样,通过定义一个词对齐集合
$
S
$
,使得每次训练迭代都在
$
S
$
上进行,进而降低运算量。模型4 中
$
S
$
的定义为:
\begin{eqnarray}
S
&
=
&
N(
\tilde
{
b
}^{
\infty
}
(V(
\seq
{
s
}
|
\seq
{
t
}
;2)))
\cup
(
\mathop
{
\cup
}
\limits
_{
ij
}
N(
\tilde
{
b
}_{
i
\leftrightarrow
j
}^{
\infty
}
(V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
,
2))))
S
&
=
&
N(
\tilde
{
b
}^{
\infty
}
(V(
\seq
{
s
}
|
\seq
{
t
}
;2)))
\cup
(
\mathop
{
\cup
}
\limits
_{
ij
}
N(
\tilde
{
b
}_{
i
\leftrightarrow
j
}^{
\infty
}
(V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
;
2))))
\label
{
eq:1.22
}
\end{eqnarray}
...
...
@@ -435,7 +435,7 @@ z_{>1}(\Delta_j|cb,v;\seq{a},\seq{s},\seq{t}) & = & \sum_{i=1}^l\Big[\varepsilon
\parinterval
在模型5中同样需要定义一个词对齐集合
$
S
$
,使得每次迭代都在
$
S
$
上进行。可以对
$
S
$
进行如下定义
\begin{eqnarray}
S
&
=
&
N(
\tilde
{
\tilde
{
b
}}^{
\infty
}
(V(
\seq
{
s
}
|
\seq
{
t
}
;2)))
\cup
(
\mathop
{
\cup
}
\limits
_{
ij
}
N(
\tilde
{
\tilde
{
b
}}_{
i
\leftrightarrow
j
}^{
\infty
}
(V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
,
2))))
S
&
=
&
N(
\tilde
{
\tilde
{
b
}}^{
\infty
}
(V(
\seq
{
s
}
|
\seq
{
t
}
;2)))
\cup
(
\mathop
{
\cup
}
\limits
_{
ij
}
N(
\tilde
{
\tilde
{
b
}}_{
i
\leftrightarrow
j
}^{
\infty
}
(V
_{
i
\leftrightarrow
j
}
(
\seq
{
s
}
|
\seq
{
t
}
;
2))))
\label
{
eq:1.29
}
\end{eqnarray}
\vspace
{
0.5em
}
...
...
ChapterPostscript/postscript.tex
查看文件 @
30d2c4db
差异被折叠。
点击展开。
ChapterPreface/chapterpreface.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
mt-book-xelatex.tex
查看文件 @
30d2c4db
...
...
@@ -2,8 +2,8 @@
% !TEX encoding = UTF-8 Unicode
%----------------------------------------------------------------------------------------
% 机器翻译:
统计建模与深度学习方法
% Machine Translation:
Statistical Modeling and Deep Learning Method
s
% 机器翻译:
基础与模型
% Machine Translation:
Foundations and Model
s
%
% Copyright 2020
% 肖桐(xiaotong@mail.neu.edu.cn) 朱靖波 (zhujingbo@mail.neu.edu.cn)
...
...
@@ -105,9 +105,9 @@
\thispagestyle
{
empty
}
{
\large
\noindent
{
\color
{
red
}
在此感谢为本书做出贡献的
小牛团队(部分)成员
}
\\
\noindent
{
\color
{
red
}
在此感谢为本书做出贡献的
人
}
\\
\noindent
曹润柘、曾信、孟霞、单韦乔、周涛、周书含、许诺、李北、许晨、林野、李垠桥、王子扬、刘辉、张裕浩、冯凯、罗应峰、魏冰浩、王屹超、李炎洋、
姜雨帆、田丰宁、刘继强、张哲旸、陈贺轩、刘晓倩、牛蕊、杜权、胡驰、王泽洋、刘腾博、刘兴宇、徐萍、赵闯、高博、张春良、王会珍、张俐、杨木润、宁义明、李洋、秦浩、胡明涵、马安香
\\
\noindent
曹润柘、曾信、孟霞、单韦乔、周涛、周书含、许诺、李北、许晨、林野、李垠桥、王子扬、刘辉、张裕浩、冯凯、罗应峰、魏冰浩、王屹超、李炎洋、
胡驰、姜雨帆、田丰宁、刘继强、张哲旸、陈贺轩、牛蕊、杜权、张春良、王会珍、张俐、马安香、胡明涵
\\
}
...
...
@@ -144,14 +144,14 @@
%\include{Chapter10/chapter10}
%\include{Chapter11/chapter11}
%\include{Chapter12/chapter12}
%
\include{Chapter13/chapter13}
\include
{
Chapter13/chapter13
}
%\include{Chapter14/chapter14}
%\include{Chapter15/chapter15}
%\include{Chapter16/chapter16}
%\include{Chapter17/chapter17}
%\include{Chapter18/chapter18}
\include
{
ChapterPostscript/postscript
}
\include
{
ChapterAcknowledgement/acknowledgement
}
%
\include{ChapterPostscript/postscript}
%
\include{ChapterAcknowledgement/acknowledgement}
%\include{ChapterAppend/chapterappend}
...
...
编写
预览
Markdown
格式
0%
重试
或
添加新文件
添加附件
取消
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!
取消
请
注册
或者
登录
后发表评论