Merge pull request #621 from karlnapf/master

HSIC statistic, fixes and new interfaces
shogun-toolbox · Jul 4, 2012 · caa554c · caa554c
2 parents af1733e + 0fc3472
commit caa554c
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 3 deletions.
diff --git a/src/interfaces/modular/Statistics.i b/src/interfaces/modular/Statistics.i
@@ -13,10 +13,16 @@
 %rename(KernelTwoSampleTestStatistic) CKernelTwoSampleTestStatistic;
 %rename(LinearTimeMMD) CLinearTimeMMD;
 %rename(QuadraticTimeMMD) CQuadraticTimeMMD;
+%rename(IndependenceTestStatistic) CIndependenceTestStatistic;
+%rename(KernelIndependenceTestStatistic) CKernelIndependenceTestStatistic;
+%rename(HSIC) CHSIC;
 
 /* Include Class Headers to make them visible from within the target language */
 %include <shogun/statistics/TestStatistic.h>
 %include <shogun/statistics/TwoSampleTestStatistic.h>
 %include <shogun/statistics/KernelTwoSampleTestStatistic.h>
 %include <shogun/statistics/LinearTimeMMD.h>
 %include <shogun/statistics/QuadraticTimeMMD.h>
+%include <shogun/statistics/IndependenceTestStatistic.h>
+%include <shogun/statistics/KernelIndependenceTestStatistic.h>
+%include <shogun/statistics/HSIC.h>
diff --git a/src/interfaces/modular/Statistics_includes.i b/src/interfaces/modular/Statistics_includes.i
@@ -4,5 +4,8 @@
  #include <shogun/statistics/KernelTwoSampleTestStatistic.h>
  #include <shogun/statistics/LinearTimeMMD.h>
  #include <shogun/statistics/QuadraticTimeMMD.h>
+ #include <shogun/statistics/IndependenceTestStatistic.h>
+ #include <shogun/statistics/KernelIndependenceTestStatistic.h>
+ #include <shogun/statistics/HSIC.h>
 %}
 
diff --git a/src/shogun/statistics/HSIC.cpp b/src/shogun/statistics/HSIC.cpp
@@ -22,6 +22,12 @@ CHSIC::CHSIC() : CKernelIndependenceTestStatistic()
 CHSIC::CHSIC(CKernel* kernel_p, CKernel* kernel_q, CFeatures* p,CFeatures* q) :
 		CKernelIndependenceTestStatistic(kernel_p, kernel_q, p, q)
 {
+	if (p->get_num_vectors()!=q->get_num_vectors())
+	{
+		SG_ERROR("%s: Only features with equal number of vectors "
+				"are currently possible\n", get_name());
+	}
+
 	init();
 }
 
@@ -44,7 +50,29 @@ float64_t CHSIC::compute_statistic()
 				get_name());
 	}
 
-	return 0;
+	/* compute kernel matrices (these have to be stored unfortunately */
+	m_kernel_p->init(m_p, m_p);
+	m_kernel_q->init(m_q, m_q);
+
+	SGMatrix<float64_t> K=m_kernel_p->get_kernel_matrix();
+	SGMatrix<float64_t> L=m_kernel_p->get_kernel_matrix();
+
+	/* center matrices (replaces this H matrix from the paper) */
+	K.center();
+	L.center();
+
+	/* compute MATLAB: sum(sum((H*K)' .* (H*L))), which is biased HSIC */
+	index_t m=K.num_rows;
+	float64_t result=0;
+	for (index_t i=0; i<m; ++i)
+	{
+		for (index_t j=0; j<m; ++j)
+			result+=K(j,i)*L(i,j);
+	}
+
+	result/=m*m;
+
+	return result;
 }
 
 float64_t CHSIC::compute_p_value(float64_t statistic)

diff --git a/src/shogun/statistics/HSIC.h b/src/shogun/statistics/HSIC.h
@@ -18,7 +18,7 @@ namespace shogun
 /** TODO
  *
  */
-class CHSIC : CKernelIndependenceTestStatistic
+class CHSIC : public CKernelIndependenceTestStatistic
 {
 public:
 	/** TODO */
@@ -35,7 +35,7 @@ class CHSIC : CKernelIndependenceTestStatistic
 
 	virtual ~CHSIC();
 
-	/** TODO */
+	/** Computes the biased HSIC TODO */
 	virtual float64_t compute_statistic();
 
 	/** computes a p-value based on current method for approximating the