# Author: Alexander Fabisch <afabisch@informatik.uni-bremen.de>
import numpy as np
from .optimizer import Optimizer
from ..utils.validation import check_random_state
[docs]class NoOptimizer(Optimizer):
"""No optimizer.
Implements the optimizer interface but does not modify any parameters.
Parameters
----------
initial_params : array-like, shape = (n_params,), optional (default: [0, 0])
Initial parameter vector.
"""
[docs] def __init__(self, initial_params=None, **kwargs):
self.initial_params = initial_params
def init(self, dimension):
if self.initial_params is None:
self.initial_params = np.zeros(dimension)
else:
self.initial_params = np.asarray(self.initial_params).astype(
np.float64, copy=True)
if dimension != len(self.initial_params):
raise ValueError("Number of dimensions (%d) does not match "
"number of initial parameters (%d)."
% (dimension, len(self.initial_params)))
self.best_params = np.asarray(self.initial_params).copy()
def get_next_parameters(self, p, explore=True):
p[:] = self.best_params
def set_evaluation_feedback(self, rewards):
pass
def get_best_parameters(self):
return self.best_params
def is_behavior_learning_done(self):
return False
[docs]class RandomOptimizer(Optimizer):
"""Random optimizer.
Parameters
----------
initial_params : array-like, shape = (n_params,), optional (default: [0, 0])
Initial parameter vector.
covariance : array-like, shape = (n_params,), optional (default: I)
Exploration covariance.
random_state : int, optional
Seed for the random number generator.
"""
[docs] def __init__(self, initial_params=None, covariance=None, random_state=None,
**kwargs):
self.initial_params = initial_params
self.covariance = covariance
self.random_state = random_state
def init(self, dimension):
if self.initial_params is None:
self.initial_params = np.zeros(dimension)
else:
self.initial_params = np.asarray(self.initial_params).astype(
np.float64, copy=True)
if dimension != len(self.initial_params):
raise ValueError("Number of dimensions (%d) does not match "
"number of initial parameters (%d)."
% (dimension, len(self.initial_params)))
self.best_params = np.asarray(self.initial_params).copy()
self.params = np.zeros(dimension)
if self.covariance is None:
self.covariance = np.eye(dimension)
self.random_state = check_random_state(self.random_state)
self.best_reward = -np.inf
def get_next_parameters(self, p):
self.params = self.random_state.multivariate_normal(
self.initial_params, self.covariance, size=1)[0]
p[:] = self.params
def set_evaluation_feedback(self, rewards):
r = np.sum(rewards)
if r > self.best_reward:
self.best_reward = r
self.best_params = self.params.copy()
def get_best_parameters(self):
return self.best_params
def is_behavior_learning_done(self):
return False
