1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
|
// This is mul/mbl/mbl_rvm_regression_builder.h
#ifndef mbl_rvm_regression_builder_h_
#define mbl_rvm_regression_builder_h_
//:
// \file
// \brief Object to train Relevance Vector Machines for regression
// \author Tim Cootes
#include <vcl_vector.h>
#include <vnl/vnl_vector.h>
#include <vnl/vnl_matrix.h>
#include <mbl/mbl_data_wrapper.h>
//=======================================================================
//: Object to train Relevance Vector Machines for regression.
// Trains Relevance Vector Machines (see papers by Michael Tipping)
// for regression.
class mbl_rvm_regression_builder
{
private:
//: Record of mean weights
vnl_vector<double> mean_wts_;
//: Record of covariance in weights
vnl_matrix<double> S_;
//: Compute design matrix F from subset of elements in kernel matrix
// Uses gaussian distance expression with variance var
void design_matrix(const vnl_matrix<double>& kernel_matrix,
const vcl_vector<int>& index,
vnl_matrix<double>& F);
//: Perform one iteration of optimisation
bool update_step(const vnl_matrix<double>& design_matrix,
const vnl_vector<double>& targets,
const vcl_vector<int>& index0,
const vcl_vector<double>& alpha0,
double sqr_width0,
vcl_vector<int>& index,
vcl_vector<double>& alpha,
double &error_var);
public:
//: Dflt ctor
mbl_rvm_regression_builder();
//: Destructor
virtual ~mbl_rvm_regression_builder();
//: Train RVM given a set of vectors and set of target values
// Resulting RVM has form f(x)=w[0]+sum w[i+1]K(x,data[index[i]])
// where K(x,y)=exp(-|x-y|^2/2var), and index.size() gives
// the number of the selected vectors.
// Note that on exit, weights.size()==index.size()+1
// weights[0] is the constant offset, and weights[i+1]
// corresponds to selected input vector index[i].
// \param data[i] training vectors
// \param targets[i] gives value at vector i
// \param index returns indices of selected vectors
// \param weights returns weights for selected vectors
// \param error_var returns estimated error variance for resulting function
void gauss_build(mbl_data_wrapper<vnl_vector<double> >& data,
double var,
const vnl_vector<double>& targets,
vcl_vector<int>& index,
vnl_vector<double>& weights,
double &error_var);
//: Train RVM given a distance matrix and set of target values
// Resulting RVM has form f(x)=w[0]+sum w[i+1]K(x,vec[index[i]])
// where K(x,y) is the kernel function, and index.size() gives
// the number of the selected vectors.
// Assuming the original data is vec[i], then on input we should
// have kernel_matrix(i,j)=K(vec[i],vec[j])
// Note that on exit, weights.size()==index.size()+1
// weights[0] is the constant offset, and weights[i+1]
// corresponds to selected input vector index[i].
//
// The algorithm involves inverting an (n+1)x(n+1) matrix,
// where n is the number of vectors to consider as relevant
// vectors. This doesn't have to be all the samples.
// For efficiency, one can provide the kernel matrix as an
// m x n matrix, where m is the number of samples to test
// against (targets.size()==m) but n<=m samples are to be
// considered as potential relevant vectors (ie the first n).
// \param kernel_matrix (i,j) element gives kernel function between i and j training vectors
// \param targets[i] gives value at vector i
// \param index returns indices of selected vectors
// \param weights returns weights for selected vectors
// \param error_var returns estimated error variance for resulting function
void build(const vnl_matrix<double>& kernel_matrix,
const vnl_vector<double>& targets,
vcl_vector<int>& index,
vnl_vector<double>& weights,
double &error_var);
};
#endif
|
