?? tune_ocr.html
字號:
<html><head> <meta HTTP-EQUIV="Content-Type" CONTENT="text/html;charset=ISO-8859-1"> <title>tune_ocr.m</title><link rel="stylesheet" type="text/css" href="../../../m-syntax.css"></head><body><code><span class=h1>% TUNE_OCR Tunes SVM classifier for OCR problem.
</span><br><span class=help>%
</span><br><span class=help>% <span class=help_field>Description:</span></span><br><span class=help>% The following steps are performed:
</span><br><span class=help>% - Training set is created from data in directory ExamplesDir.
</span><br><span class=help>% - Multi-class SVM is trained for a set of arguments and
</span><br><span class=help>% regularization constants. The best model is selected
</span><br><span class=help>% based on the cross-validation error.
</span><br><span class=help>%
</span><br><hr><br><span class=help1>% <span class=help1_field>(c)</span> Statistical Pattern Recognition Toolbox, (C) 1999-2003,
</span><br><span class=help1>% Written by Vojtech Franc and Vaclav Hlavac,
</span><br><span class=help1>% <a href="http://www.cvut.cz">Czech Technical University Prague</a>,
</span><br><span class=help1>% <a href="http://www.feld.cvut.cz">Faculty of Electrical engineering</a>,
</span><br><span class=help1>% <a href="http://cmp.felk.cvut.cz">Center for Machine Perception</a>
</span><br><br><span class=help1>% <span class=help1_field>Modifications:</span>
</span><br><span class=help1>% 04-jun-2004, VF
</span><br><span class=help1>% 09-sep-2003, VF
</span><br><br><hr>cd /home.dokt/xfrancv/work/new_stprtool/;
<br>stprpath;
<br>cd /home.dokt/xfrancv/work/new_stprtool/demos/ocr;
<br>
<br><span class=comment>% Setting
</span><br><span class=comment>%===================================
</span><br>ExamplesDir = <span class=quotes>'../../data/ocr_numerals/'</span>; <span class=comment>% input directory with exmaples
</span><br>OCRTuningFileName = <span class=quotes>'ocrtuning'</span>; <span class=comment>% output file with result of tuning
</span><br>
<br><span class=comment>% Model setting for multi-class SVM
</span><br>options.ker = <span class=quotes>'rbf'</span>; <span class=comment>% kernel type
</span><br>options.arg = [1 2.5 5 7.5 10] ; <span class=comment>% kernel argument
</span><br>options.C = [inf]; <span class=comment>% regularization constant
</span><br>options.verb = 1; <span class=comment>% display progress info
</span><br>options.solver =<span class=quotes>'oaosvm'</span>;
<br>options.num_fold = 5;
<br>options.svm_options.solver = <span class=quotes>'svmlight'</span>;
<br>
<br><span class=comment>% Create training set
</span><br><span class=comment>%====================================
</span><br><span class=io>fprintf</span>(<span class=quotes>'Creating training set:\n'</span>);
<br>TrainingDataFile = [ExamplesDir <span class=quotes>'OcrTrndata.mat'</span>];
<br>mergesets( ExamplesDir, TrainingDataFile );
<br>data = load(TrainingDataFile );
<br>
<br><span class=comment>% Tuning SVM model
</span><br><span class=comment>%====================================
</span><br>
<br><span class=io>fprintf</span>(<span class=quotes>'Tuning multi-class SVM classifier.\n'</span>);
<br>[model,Error] = evalsvm(data,options);
<br>
<br><span class=io>fprintf</span>(<span class=quotes>'\nSaving results to: %s\n'</span>,OCRTuningFileName);
<br>save(OCRTuningFileName,<span class=quotes>'model'</span>,<span class=quotes>'Error'</span>,<span class=quotes>'options'</span>);
<br>
<br><span class=comment>% Visualization
</span><br><span class=comment>%====================================
</span><br><span class=comment>%figure;
</span><br><span class=comment>%load(OCRTuningFileName)
</span><br><span class=comment>%figure; mesh(options.arg,options.C,Errors);
</span><br><span class=comment>%hold on; xlabel('arg'); ylabel('C');
</span><br><span class=comment>%figure; contour(options.arg,options.C,Errors);
</span><br><span class=comment>%hold on; xlabel('arg'); ylabel('C');
</span><br>
<br><span class=comment>% EOF</span><br></code>?? 快捷鍵說明
復制代碼
Ctrl + C
搜索代碼
Ctrl + F
全屏模式
F11
切換主題
Ctrl + Shift + D
顯示快捷鍵
?
增大字號
Ctrl + =
減小字號
Ctrl + -