karmedbanditExp3_1 Class Reference#include <karmedbandit.h>
List of all members.
Detailed Description
Makes decisions regarding an adversarial k-armed bandit.
Uses algorithms described in: The non-stochastic multi-armed bandit problem Auer, Cesa-Bianchi, Freund, and Schapire October 14, 2002
Definition at line 84 of file karmedbandit.h.
|
Public Member Functions |
| | karmedbanditExp3_1 (unsigned int k) |
| | constructor, pass the number of arms
|
| unsigned int | decide () |
| | returns the next choice, [0:k-1]
|
| void | reward (bool rew) |
| | call this if you want to reward (r==true) or penalize (r==false) the previous decision
|
Protected Member Functions |
| void | restart () |
| | restarts exp3
|
Protected Attributes |
| unsigned int | r |
| | the number of restarts
|
| double | gr |
| | the gamma_r parameter
|
| unsigned int | last |
| | the last choice
|
| std::vector< double > | G |
| | the G-hat's
|
| karmedbanditExp3 | exp3 |
| | runs exp3 within this
|
Constructor & Destructor Documentation
| karmedbanditExp3_1::karmedbanditExp3_1 |
( |
unsigned int |
k |
) |
[inline] |
|
|
|
constructor, pass the number of arms
Definition at line 87 of file karmedbandit.h. |
Member Function Documentation
| unsigned int karmedbanditExp3_1::decide |
( |
|
) |
[inline] |
|
| void karmedbanditExp3_1::restart |
( |
|
) |
[inline, protected] |
|
| void karmedbanditExp3_1::reward |
( |
bool |
rew |
) |
[inline] |
|
Member Data Documentation
The documentation for this class was generated from the following file:
|