From a11a8e15c71f711eb0dbef17b6c651557d0c75ef Mon Sep 17 00:00:00 2001 From: "philblunsom@gmail.com" Date: Wed, 28 Jul 2010 21:05:31 +0000 Subject: Presentation given, missing closing. git-svn-id: https://ws10smt.googlecode.com/svn/trunk@457 ec762483-ff6d-05da-a07a-a48fb63a330f --- report/intro_slides/final_slides.pdf | Bin 1926728 -> 1826408 bytes report/intro_slides/final_slides.tex | 118 ++++++++++------------------------- 2 files changed, 34 insertions(+), 84 deletions(-) diff --git a/report/intro_slides/final_slides.pdf b/report/intro_slides/final_slides.pdf index e43a6d82..8d67cd20 100644 Binary files a/report/intro_slides/final_slides.pdf and b/report/intro_slides/final_slides.pdf differ diff --git a/report/intro_slides/final_slides.tex b/report/intro_slides/final_slides.tex index f552a8b4..b37348b5 100644 --- a/report/intro_slides/final_slides.tex +++ b/report/intro_slides/final_slides.tex @@ -299,7 +299,7 @@ \end{exampleblock} \begin{itemize} \only<1>{\item S -> $\langle$ NP\ind{1} VP\ind{2}, NP\ind{1} VP\ind{2} $\rangle$, \\ NP -> $\langle$ PRP\ind{1}, PRP\ind{1} $\rangle$} -\only<1>{\item PRP -> $\langle$ Je, I $\rangle$, \\ VP -> $\langle$ ne veux pas VB\ind{1}, do not want to VB\ind{1} $\rangle$, \\ VB -> $\langle$ travailler, work $\rangle$} +\only<1>{\item PRP -> $\langle$ Je, I $\rangle$, VB -> $\langle$ travailler, work $\rangle$ \\ VP -> $\langle$ ne veux pas VB\ind{1}, do not want to VB\ind{1} $\rangle$} \only<2->{ \item Strong model of sentence structure. \item Reliant on a treebank to train the parser. @@ -471,13 +471,23 @@ Output: \begin{frame}[t]{Workshop Streams} -Expand, describing challenges faced in each stream. -\vspace{0.25in} -\begin{unpacked_itemize} -\item Implement scalable SCFG grammar extraction algorithms. -\item Improve SCFG decoders to efficiently handle the grammars produce. -\item Investigate discriminative training regimes to leverage features extracted from these grammars. -\end{unpacked_itemize} +\begin{enumerate} +\item Implement scalable labelled SCFG grammar induction algorithms: +\begin{itemize} +\item by clustering translation phrases which occur in the same context we can learn which phrases are substituteable, +\item we have implemented both parametric and non-parametric Bayesian clustering algorithms. +\end{itemize} +\item Improve SCFG decoders to efficiently handle the grammars produced: +\begin{itemize} +\item translation complexity scales quadratically as we add more categories, +\item in order to decode efficiently with the grammars we've induced we have created faster search algorithms tuned for syntactic grammars. +\end{itemize} +\item Investigate discriminative training regimes to leverage features extracted from these grammars: +\begin{itemize} +\item to make the most of our induced grammars we need discriminative training algorithms that learn from more than a handful of features, +\item we've implemented two large scale discriminative algorithms for training our models. +\end{itemize} +\end{enumerate} \end{frame} \begin{frame}[t]{Extrinsic evaluation: Bleu} @@ -565,7 +575,7 @@ Expand, describing challenges faced in each stream. \item Europarl Dutch-French: \begin{itemize} \item 100k sentence pairs, standard Europarl test sets - \item Hiero baseline score: Europarl 2008 - 26.3 (1 reference) + \item Hiero baseline score: Europarl 2008 - 15.75 (1 reference) \item Major challenges: V2 / V-final word order, morphology \end{itemize} \end{unpacked_itemize} @@ -595,11 +605,11 @@ Expand, describing challenges faced in each stream. \end{figure} \end{exampleblock} - \vspace{0.25in} + %\vspace{0.25in} \end{column} \begin{column}{0.7\textwidth} \begin{unpacked_itemize} - \item 1:55pm Experimental Setup. Trevor + \item 1:55pm Grammar induction and evaluation. Trevor \item 2:10pm Non-parametric models of category induction. Chris \item 2:25pm Inducing categories for morphology. Jan \item 2:35pm Smoothing, backoff and hierarchical grammars. Olivia @@ -635,7 +645,6 @@ Expand, describing challenges faced in each stream. \end{figure} \end{exampleblock} - \vspace{0.25in} \end{column} \begin{column}{0.7\textwidth} \begin{itemize} @@ -661,8 +670,6 @@ Expand, describing challenges faced in each stream. \end{unpacked_itemize} \end{frame} -\begin{frame}[t]{This slide is intentionally left blank.} -\end{frame} \begin{frame}[t]{Outline} \begin{columns} @@ -687,12 +694,12 @@ Expand, describing challenges faced in each stream. \end{figure} \end{exampleblock} - \vspace{0.25in} + %\vspace{0.25in} \end{column} \begin{column}{0.7\textwidth} \begin{unpacked_itemize} - \only<1>{\item \alert{1:55pm Motivation and experimental methodology. Trevor}} - \only<2->{\item 1:55pm Motivation and experimental methodology. Trevor} + \only<1>{\item \alert{1:55pm Grammar induction and evaluation. Trevor}} + \only<2->{\item 1:55pm Grammar induction and evaluation. Trevor} \only<2>{\item \alert{2:10pm Non-parametric models of category induction. Chris}} \only<1,3->{\item 2:10pm Non-parametric models of category induction. Chris} \only<3>{\item \alert{2:25pm Inducing categories for morphology. Jan}} @@ -733,83 +740,26 @@ Expand, describing challenges faced in each stream. \end{figure} \end{exampleblock} - \vspace{0.25in} \end{column} \begin{column}{0.7\textwidth} \begin{itemize} \setlength{\itemsep}{25pt} \setlength{\parskip}{0pt} \setlength{\parsep}{0pt} - \only<1>{\item \alert{3:15pm Training models with rich features spaces. Vlad}} - \only<2->{\item 3:15pm Training models with rich features spaces. Vlad} - \only<2>{\item \alert{3:30pm Decoding with complex grammars. Adam}} - \only<1,3->{\item 3:30pm Decoding with complex grammars. Adam} - \only<3>{\item \alert{4:00pm Closing remarks. Phil}} - \only<1,2,4->{\item 4:00pm Closing remarks. Phil} - \only<4>{\item \alert{4:05pm Finish.}} - \only<1-3>{\item 4:05pm Finish.} + \only<1>{\item \alert{3:20pm Parametric models: posterior regularisation. Desai}} + \only<2->{\item 3:20pm Parametric models: posterior regularisation. Desai} + \only<2>{\item \alert{3:35pm Training models with rich features spaces. Vlad}} + \only<1,3->{\item 3:35pm Training models with rich features spaces. Vlad} + \only<3>{\item \alert{3:50pm Decoding with complex grammars. Adam}} + \only<1,2,4->{\item 3:50pm Decoding with complex grammars. Adam} + \only<4>{\item \alert{4:20pm Closing remarks. Phil}} + \only<1-3,5->{\item 4:20pm Closing remarks. Phil} + \only<5>{\item \alert{4:25pm Finish.}} + \only<1-4>{\item 4:25pm Finish.} \end{itemize} \end{column} \end{columns} \end{frame} -\begin{frame}[t]{This slide is intentionally left blank.} -\end{frame} - - -\begin{frame}[t]{Statistical machine translation: state-of-the-art} -%\vspace{1.0cm} -\begin{exampleblock}{Urdu $\rightarrow$ English} - \begin{figure} - {\centering \includegraphics[scale=0.55]{urdu-bl.pdf}} - \end{figure} -\end{exampleblock} -\begin{itemize} - \item Current state-of-the-art translation models struggle with language pairs which exhibit large differences in structure. -\end{itemize} -\end{frame} - -\begin{frame}[t]{Statistical machine translation: our unsupervised grammars} -%\vspace{1.0cm} -\begin{exampleblock}{Urdu $\rightarrow$ English} - \begin{figure} - {\centering \includegraphics[scale=0.55]{urdu-25hp.pdf}} - \end{figure} -\end{exampleblock} -\begin{itemize} - \item In this workshop we've made some small steps towards better translations for difficult language pairs. -\end{itemize} -\only<2->{ - Google Translate: \\ - \hspace{0.25in} {\em *After the attack a number of local residents has blank areas.} -} -\end{frame} - - -\begin{frame}[t]{Induced Translation Structure} -\begin{center} -\includegraphics[scale=0.32]{joshua_tree19.pdf} -\end{center} -\end{frame} - -\begin{frame}[t]{What we've achieved:} - \vspace{0.5in} - \begin{unpacked_itemize} - \item - \item - \end{unpacked_itemize} -\end{frame} - - -\begin{frame}[t]{We're we'll go from here:} - \vspace{0.5in} - \begin{unpacked_itemize} - \item - \item - \end{unpacked_itemize} -\end{frame} - - - \end{document} -- cgit v1.2.3