FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads > Class Template Reference

Implements threaded version of randomized search that repeatedly samples random subsets to eventually yield the one with highest criterion value. More...

#include <search_monte_carlo_threaded.hpp>

Inheritance diagram for FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >:

[legend]

Collaboration diagram for FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >:

[legend]

List of all members.

Classes
struct	ThreadLocal
	thread-local storage of current subset candidate, criterion clone, and tracker clone More...
Public Types
typedef Search< RETURNTYPE, DIMTYPE, SUBSET, CRITERION >	parent
typedef boost::shared_ptr < CRITERION >	PCriterion
typedef boost::shared_ptr< SUBSET >	PSubset
Public Member Functions
	Search_Monte_Carlo_Threaded (const unsigned long trials_limit=1000, const unsigned long time_limit=0, const unsigned int time_limit_check_freq=10)
virtual bool	search (const DIMTYPE target_d, RETURNTYPE &result, const PSubset sub, const PCriterion crit, std::ostream &os=std::cout)
	returns found subset of target_d features (optimizes cardinality if target_d==0) + criterion value
void	set_stopping_condition (const unsigned long trials_limit=1000, const unsigned long time_limit=0, const unsigned int time_limit_check_freq=100)
	set maximum number of random trials or maximum time in seconds or both
void	set_cardinality_randomization (const DIMTYPE d_from, const DIMTYPE d_to)
	generate subset size first from [d_from,d_to], then randomly choose exactly that number of features
void	set_cardinality_randomization (const float d_prob)
	for each feature roll the dice and select it with probability d_prob
void	reset_cardinality_randomization ()
	equivalent to set_cardinality_randomization(1,no_of_all_features)
virtual std::ostream &	print (std::ostream &os) const
Protected Member Functions
void	evaluator_thread (unsigned int idx)
unsigned long	next_trial ()
Protected Attributes
PCriterion	template_crit
ThreadLocal	tlocal [max_threads]
boost::mutex	mutex_trials_out
boost::mutex	mutex_random_sub
boost::shared_ptr< StopWatch >	swatch
unsigned long	_trials
	to be updated concurrently from threads
volatile DIMTYPE	_target_d
unsigned long	_time_limit
	max time allowed to be spent in seconds
unsigned long	_trials_limit
	max no of candidates evaluated
unsigned int	_time_limit_check_freq
	number of trials before next time check
DIMTYPE	_d_from
DIMTYPE	_d_to
	generate random subsets of size within these limits
float	_d_prob
	active d_prob must be from (0,1]; generates random subset of random size where the mean size over a series of calls is d_prob*no_of_all_features

---

Detailed Description

template<class RETURNTYPE, typename DIMTYPE, class SUBSET, class CRITERION, unsigned int max_threads = 2>
class FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >

Implements threaded version of randomized search that repeatedly samples random subsets to eventually yield the one with highest criterion value.

Concurrently evaluates each subset candidate using clones of the supplied criterion and returns the one subset that yielded highest criterion value

Note:

Requires user-set terminating condition, either in form of the number of random trials accomplished or in form of maximum time spent or both.

offers two ways of random subset generation: a) set_cardinality_randomization(const DIMTYPE d_from, const DIMTYPE d_to) invokes the first option, i.e., when generating a random subset, the subset size is first chosen randomly from [d_from,d_to], then features are randomly selected until the required number is reached b) set_cardinality_randomization(const float d_prob) invokes the second option, where each feature is randomly included/excluded with probability d_prob. The expected subset size is thus d_prob*no_of_all_features, but can be actually anything from [0,no_of_all_features].

Examples:

demo24t.cpp, and demo35t.cpp.

---

Member Function Documentation

template<class RETURNTYPE, typename DIMTYPE, class SUBSET, class CRITERION, unsigned int max_threads = 2>

unsigned long FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >::next_trial

(

)

[inline, protected]

< (synchronized) _trials increase + test whether _trials_limit has been reached

References FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >::_trials.

---

Member Data Documentation

template<class RETURNTYPE, typename DIMTYPE, class SUBSET, class CRITERION, unsigned int max_threads = 2>

PCriterion FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >::template_crit [protected]

to detect change of criterion inbetween search() calls

Referenced by FST::Search_Monte_Carlo_Threaded< RETURNTYPE, DIMTYPE, SUBSET, CRITERION, max_threads >::search().

---

The documentation for this class was generated from the following file:

• search_monte_carlo_threaded.hpp