PLearn 0.1
|
description of the main class: TransformationLearner More...
#include <TransformationLearner.h>
Public Member Functions | |
ReconstructionCandidate (int targetIdx_=-1, int neighborIdx_=-1, int transformIdx_=-1, real weight_=0) | |
Public Attributes | |
int | targetIdx |
int | neighborIdx |
int | transformIdx |
real | weight |
description of the main class: TransformationLearner
GENERATION PROCESS
We suppose a new point v is obtained from a point x by : 1) choosing a transformation t among a set of transformations T (with probability p(t)) 2) applying the choosen transformation t on x 3) add some noise in all the directions (noise = normally distributed random variable)
UNDERLYING PROBABILITY SPACE
variables of the distribution : X : real vector V : real vector, neighbor of X T : transformation P(x,v,t) = P(v is obtained by applying transformation t on x) = P(x,v|t)P(t) = N( T(t)(x)|v,sigma)P(t)
LEARNING BEHAVIOR
The parameters of the distributions are learned using a variant of E.M.algorithm
Definition at line 119 of file TransformationLearner.h.
PLearn::ReconstructionCandidate::ReconstructionCandidate | ( | int | targetIdx_ = -1 , |
int | neighborIdx_ = -1 , |
||
int | transformIdx_ = -1 , |
||
real | weight_ = 0 |
||
) | [inline] |
Definition at line 125 of file TransformationLearner.h.
References neighborIdx, targetIdx, transformIdx, and weight.
{ targetIdx = targetIdx_; neighborIdx = neighborIdx_; transformIdx = transformIdx_ ; weight = weight_; }
Definition at line 122 of file TransformationLearner.h.
Referenced by PLearn::TransformationLearner::computeReconstructionWeight(), PLearn::operator<<(), PLearn::operator>>(), and ReconstructionCandidate().
Definition at line 122 of file TransformationLearner.h.
Referenced by PLearn::TransformationLearner::computeReconstructionWeight(), PLearn::operator<<(), PLearn::operator>>(), and ReconstructionCandidate().
Definition at line 122 of file TransformationLearner.h.
Referenced by PLearn::TransformationLearner::computeReconstructionWeight(), PLearn::operator<<(), PLearn::operator>>(), and ReconstructionCandidate().
Definition at line 123 of file TransformationLearner.h.
Referenced by PLearn::operator<(), PLearn::operator<<(), PLearn::operator==(), PLearn::operator>>(), and ReconstructionCandidate().