📄 svm_struct.html
字号:
<HTML><HEAD><META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=windows-1252"><META NAME="Generator" CONTENT="Microsoft FrontPage 4.0"><TITLE>SVM-Struct Support Vector Machine for Complex Outputs</TITLE><META NAME="Version" CONTENT="8.0.3514"><META NAME="Date" CONTENT="11/26/96"><META NAME="Template" CONTENT="C:\Programme\Microsoft Office\Office\HTML.DOT"></HEAD><BODY vLink=#800080 link=#0000ff bgColor=#ffffff><TABLE cellSpacing=0 cellPadding=5 border=0> <TBODY> <TR> <TD vAlign=top width="14%"> <H2><IMG height=80 src="culogo_125.gif" tppabs="http://www.cs.cornell.edu/People/tj/images/culogo_125.gif" width=80></H2></TD> <TD vAlign=top width="75%"> <H1 align=center>SVM<I><SUP>struct</SUP> </H1></I> <H1 align=center>Support Vector Machine for Complex Outputs</H1><FONT color=#000000> <P align=center>Author: </FONT><A href="javascript:if(confirm('http://www.joachims.org/ \n\nThis file was not retrieved by Teleport Pro, because it is addressed on a domain or path outside the boundaries set for its Starting Address. \n\nDo you want to open it from the server?'))window.location='http://www.joachims.org/'" tppabs="http://www.joachims.org/" target=_top>Thorsten Joachims</A><FONT color=#000000> <</FONT><A href="mailto:thorsten@joachims.org">thorsten@joachims.org</A><FONT color=#000000>> <BR></FONT><A href="javascript:if(confirm('http://www.cornell.edu/ \n\nThis file was not retrieved by Teleport Pro, because it is addressed on a domain or path outside the boundaries set for its Starting Address. \n\nDo you want to open it from the server?'))window.location='http://www.cornell.edu/'" tppabs="http://www.cornell.edu/" target=_top>Cornell University</A><FONT color=#000000> <BR></FONT><A href="javascript:if(confirm('http://www.cs.cornell.edu/ \n\nThis file was not retrieved by Teleport Pro, because it is addressed on a domain or path outside the boundaries set for its Starting Address. \n\nDo you want to open it from the server?'))window.location='http://www.cs.cornell.edu/'" tppabs="http://www.cs.cornell.edu/" target=_top>Department of Computer Science</A><FONT color=#000000> </P> <P align=center>Version: 2.50 <BR>Date: 20.07.2004</FONT></P></TD> <TD vAlign=top width="11%"> <H2><IMG height=80 src="culogo_125.gif" tppabs="http://www.cs.cornell.edu/People/tj/images/culogo_125.gif" width=80></H2></TD></TR></TBODY></TABLE><H2>Overview</H2><P><I>SVM<SUP>struct</SUP></I> is a Support Vector Machine (SVM) algorithm for predicting multivariate outputs. It performs supervised learning by approximating a mapping</p> <center><i>h: X --> Y</i></center><br><p>using labeled training examples <tt>(x<SUB>1</SUB>,y<SUB>1</SUB>), ..., (x<SUB>n</SUB>,y<SUB>n</SUB>)</tt>. Unlike regular SVMs, however, which consider only univariate predictions like in classification and regression, <I>SVM<SUP>struct</SUP></I> can predict complex objects <tt>y</tt> like trees, sequences, or sets. Examples of problems with complex outputs are natural language parsing, sequence alignment in protein homology detection, and markov models for part-of-speech tagging.</P><P>The sparse approximation algorithm implemented in <I>SVM<SUP>struct</SUP></I> is described in [1][2]. The implementation is based on the <a href="index.html" tppabs="http://www.cs.cornell.edu/People/tj/svm%5Flight/index.html"><I>SVM<SUP>light</SUP></I></a> quadratic optimizer [3].</P><H2>Existing Instantiations</H2><P><I>SVM<SUP>struct</SUP></I> can be thought of as an API for implementing different kinds of complex prediction algorithms. Currently, we have implemented the following learning tasks:<UL><LI><B><I>SVM<SUP>multiclass</SUP></I></B>: Multi-class classification. Learns to predict one of <tt>k</tt> mutually exclusive classes. This is probably the simplest possible instance of <I>SVM<SUP>struct</SUP></I> and serves as a tutorial example of how to use the programming interface.<BR><a href="svm_multiclass.html" tppabs="http://www.cs.cornell.edu/People/tj/svm%5Flight/svm_multiclass.html">More information and source code.</a> </LI><LI><B><I>SVM<SUP>cfg</SUP></I></B>: Learns a weighted context free grammar from examples. Training examples (e.g. for natural language parsing) specify the sentence along with the correct parse tree. The goal is to predict the parse tree of new sentences. <BR><a href="svm_cfg.html" tppabs="http://www.cs.cornell.edu/People/tj/svm%5Flight/svm_cfg.html">More information and source code.</a> </LI><LI><B><I>SVM<SUP>align</SUP></I></B>: Learning to align sequences. Given examples of how sequence pairs align, the goal is to learn the substitution matrix as well as the insertion and deletion costs of operations so that one can predict alignments of new sequences. <BR><a href="svm_align.html" tppabs="http://www.cs.cornell.edu/People/tj/svm%5Flight/svm_align.html">More information and source code.</a> </LI><LI><B><I>SVM<SUP>hmm</SUP></I></B>: Learns a Markov model from examples. Training examples (e.g. for part-of-speech tagging) specify the sequence of words along with the correct assignment of tags (i.e. states). The goal is to predict the tag sequences for new sentences. <BR>More information and source code coming soon.</LI></UL>Please let me know, if you want me to add your implementations to this list.<H2>Source Code for Implementing your Own Instantiation</H2>Instead of using one of the existing instantiations of <I>SVM<SUP>struct</SUP></I> listed above, you can implement your own. <I>SVM<SUP>struct</SUP></I> contains an API that let's you specialize the general sparse approximation training algorithm for your particular application. Referring to the algorithm as presented in [1], you merely need to provide the code for the following:<UL><LI> A function for computing the feature vector Psi.<LI> A function for computing the argmax over the (kernelized) linear discriminant function.<LI> A loss function.</UL>You can download the source code of the algorithm and the API from the following location:<P><PRE> <a href="javascript:if(confirm('http://download.joachims.org/svm_struct/current/svm_struct.tar.gz \n\nThis file was not retrieved by Teleport Pro, because it is addressed on a domain or path outside the boundaries set for its Starting Address. \n\nDo you want to open it from the server?'))window.location='http://download.joachims.org/svm_struct/current/svm_struct.tar.gz'" tppabs="http://download.joachims.org/svm_struct/current/svm_struct.tar.gz">http://download.joachims.org/svm_struct/current/svm_struct.tar.gz</a></PRE></P>The archive contains the source code of the most recent version of <I>SVM<SUP>struct</SUP></I> as well as the source code of the <I>SVM<SUP>light</SUP></I> quadratic optimizer. Unpack the archive using the shell command:<P><PRE> gunzip 朿 svm_struct.tar.gz | tar xvf ⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -