#include <p_condensation.h>
Inheritance diagram for mimas::hf::p_condensation< Hypothesis_T >:
Public Member Functions | |
void | partial_reset (const Hypothesis_T &init, float proportion, float weight_discrimination, float weight_initialization, bool drift_switch=true) |
This should be called before the filter stage occured: suppose the previous position has been identified with a reasonably good degree of confidence. | |
Protected Member Functions | |
p_condensation (std::vector< Hypothesis_T > &hypotheses) |
The same as condensation with the possibility to reinitialised part of the hypotheses.
Definition at line 15 of file p_condensation.h.
mimas::hf::p_condensation< Hypothesis_T >::p_condensation | ( | std::vector< Hypothesis_T > & | hypotheses | ) | [inline, protected] |
Definition at line 62 of file p_condensation.h.
void mimas::hf::p_condensation< Hypothesis_T >::partial_reset | ( | const Hypothesis_T & | init, | |
float | proportion, | |||
float | weight_discrimination, | |||
float | weight_initialization, | |||
bool | drift_switch = true | |||
) | [inline] |
This should be called before the filter stage occured: suppose the previous position has been identified with a reasonably good degree of confidence.
Then we can relocate around this position the previous hypotheses that had a low weight.
Another way is to look at all hypotheses and partial reinitialise around the best one.
proportion must be between 0 and 1
!!!research work to be done to automate finding of optimal parametres
Definition at line 32 of file p_condensation.h.
References Dice::uniform_random().
Here is the call graph for this function: