From b5a3dbda004c5b5330d1d8d7c67871add6ba8507 Mon Sep 17 00:00:00 2001
From: Davis King <davis@dlib.net>
Date: Fri, 31 Dec 2010 14:17:51 +0000
Subject: [PATCH] Added a SVR example program

--HG--
extra : convert_revision : svn%3Afdd8eb12-d10e-0410-9acb-85c331704f74/trunk%404061
---
 examples/CMakeLists.txt |  2 +-
 examples/svr_ex.cpp     | 88 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 89 insertions(+), 1 deletion(-)
 create mode 100644 examples/svr_ex.cpp

diff --git a/examples/CMakeLists.txt b/examples/CMakeLists.txt
index 83d5b7072..886a19544 100644
--- a/examples/CMakeLists.txt
+++ b/examples/CMakeLists.txt
@@ -84,10 +84,10 @@ add_example(surf_ex)
 add_example(svm_ex)
 add_example(svm_pegasos_ex)
 add_example(svm_sparse_ex)
+add_example(svr_ex)
 add_example(threaded_object_ex)
 add_example(thread_function_ex)
 add_example(thread_pool_ex)
 add_example(threads_ex)
 add_example(timer_ex)
 add_example(xml_parser_ex)
-
diff --git a/examples/svr_ex.cpp b/examples/svr_ex.cpp
new file mode 100644
index 000000000..b87dbd96e
--- /dev/null
+++ b/examples/svr_ex.cpp
@@ -0,0 +1,88 @@
+// The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt
+/*
+    This is an example illustrating the use of the epsilon-insensitive support vector 
+    regression object from the dlib C++ Library.
+
+    In this example we will draw some points from the sinc() function and do a
+    non-linear regression on them.
+*/
+
+#include <iostream>
+#include <vector>
+
+#include "dlib/svm.h"
+
+using namespace std;
+using namespace dlib;
+
+// Here is the sinc function we will be trying to learn with the svr_trainer 
+// object.
+double sinc(double x)
+{
+    if (x == 0)
+        return 1;
+    return sin(x)/x;
+}
+
+int main()
+{
+    // Here we declare that our samples will be 1 dimensional column vectors.  
+    typedef matrix<double,1,1> sample_type;
+
+    // Now we are making a typedef for the kind of kernel we want to use.  I picked the
+    // radial basis kernel because it only has one parameter and generally gives good
+    // results without much fiddling.
+    typedef radial_basis_kernel<sample_type> kernel_type;
+
+
+    std::vector<sample_type> samples;
+    std::vector<double> targets;
+
+    // The first thing we do is pick a few training points from the sinc() function.
+    sample_type m;
+    for (double x = -10; x <= 4; x += 1)
+    {
+        m(0) = x;
+
+        samples.push_back(m);
+        targets.push_back(sinc(x));
+    }
+
+    // Now setup a SVR trainer object.  It has three parameters, the kernel and
+    // two parameters specific to SVR.  
+    svr_trainer<kernel_type> trainer;
+    trainer.set_kernel(kernel_type(0.1));
+
+    // This parameter is the usual regularization parameter.  It determines the trade-off 
+    // between trying to reduce the training error or allowing more errors but hopefully 
+    // improving the generalization of the resulting function.  Larger values encourage exact 
+    // fitting while smaller values of C may encourage better generalization.
+    trainer.set_c(10);
+
+    // Epsilon-insensitive regression means we do regression but stop trying to fit a data 
+    // point once it is "close enough" to its target value.  This parameter is the value that 
+    // controls what we mean by "close enough".  In this case, I'm saying I'm happy if the
+    // resulting regression function gets within 0.001 of the target value.
+    trainer.set_epsilon_insensitivity(0.001);
+
+    // Now do the training and save the results
+    decision_function<kernel_type> df = trainer.train(samples, targets);
+
+    // now we output the value of the sinc function for a few test points as well as the 
+    // value predicted by SVR.
+    m(0) = 2.5; cout << sinc(m(0)) << "   " << df(m) << endl;
+    m(0) = 0.1; cout << sinc(m(0)) << "   " << df(m) << endl;
+    m(0) = -4;  cout << sinc(m(0)) << "   " << df(m) << endl;
+    m(0) = 5.0; cout << sinc(m(0)) << "   " << df(m) << endl;
+
+    // The output is as follows:
+    //  0.239389   0.23905
+    //  0.998334   0.997331
+    // -0.189201   -0.187636
+    // -0.191785   -0.218924
+
+    // The first column is the true value of the sinc function and the second
+    // column is the output from the SVR estimate.  
+}
+
+