def __init__(self,
color="black",
time_limit=5,
gui=None,
headless=False,
strategy=OthelloHeuristic.DEFAULT_STRATEGY):
super(ComputerPlayer, self).__init__(color, time_limit, gui, headless)
heuristic = OthelloHeuristic(strategy)
self.ai = GameArtificialIntelligence(heuristic.evaluate)
class ComputerPlayer(Player):
name = "ComputerPlayer"
def __init__(self,
color="black",
time_limit=5,
gui=None,
headless=False,
strategy=OthelloHeuristic.DEFAULT_STRATEGY):
super(ComputerPlayer, self).__init__(color, time_limit, gui, headless)
heuristic = OthelloHeuristic(strategy)
self.ai = GameArtificialIntelligence(heuristic.evaluate)
# Is Random For Now
def get_move(self):
import datetime
import sys
other_color = BLACK
if self.color == BLACK:
other_color = WHITE
start = datetime.datetime.now()
move = self.ai.move_search(self.current_board, self.time_limit,
self.color, other_color)
delta = datetime.datetime.now() - start
# print >> sys.stderr, "Time taken:", delta
self.apply_move(move)
return self.current_board
def __init__(self, color="black", time_limit=5, gui=None, headless=False, epochs=5, batch_size=100):
super(DeepLearningPlayer, self).__init__(color, time_limit, gui, headless)
self.model = Net()
self.ai = GameArtificialIntelligence(self.evaluate_board);
if torch.cuda.is_available():
self.model.cuda(0)
print "CUDA activated"
# print(self.model)
try:
self.model = DataHandler.load_weights(self.name)
except Exception:
if epochs != 0:
self.train_model(epochs=epochs, batch_size=batch_size)
class ComputerPlayer(Player):
def __init__(self, color="black", time_limit=5, gui=None):
super(ComputerPlayer, self).__init__(color, time_limit, gui)
heuristic = OthelloHeuristic()
self.ai = GameArtificialIntelligence(heuristic.evaluate)
# Is Random For Now
def get_move(self):
import datetime
import sys
other_color = BLACK
if self.color == BLACK:
other_color = WHITE
start = datetime.datetime.now()
move = self.ai.move_search(self.current_board, self.time_limit, self.color, other_color)
delta = datetime.datetime.now() - start
print >> sys.stderr, "Time taken:", delta
self.apply_move(move)
return self.current_board
def __init__(self, color="black", time_limit=5, gui=None):
super(ComputerPlayer, self).__init__(color, time_limit, gui)
heuristic = OthelloHeuristic()
self.ai = GameArtificialIntelligence(heuristic.evaluate)
class DeepLearningPlayer(Player):
name = "DeepLearningPlayer"
def __init__(self, color="black", time_limit=5, gui=None, headless=False, epochs=5, batch_size=100):
super(DeepLearningPlayer, self).__init__(color, time_limit, gui, headless)
self.model = Net()
self.ai = GameArtificialIntelligence(self.evaluate_board);
if torch.cuda.is_available():
self.model.cuda(0)
print "CUDA activated"
# print(self.model)
try:
self.model = DataHandler.load_weights(self.name)
except Exception:
if epochs != 0:
self.train_model(epochs=epochs, batch_size=batch_size)
def train_model(self, epochs=10, batch_size=100, continue_training=False):
losses = self.model.train_model(epochs=epochs, batch_size=batch_size, continue_training=continue_training)
DataHandler.store_weights(player_name=self.name, model=self.model)
return losses
def train_model_on_curriculum(self, epochs_per_stage=1, final_epoch=30, continue_training=False):
final_epoch = min(final_epoch, 30)
losses = self.model.train_model_on_curriculum(epochs_per_stage=epochs_per_stage, final_epoch=final_epoch, continue_training=continue_training)
DataHandler.store_weights(player_name=self.name + "_curriculum", model=self.model)
return losses
def evaluate_board(self, board, color, other_player):
sample = FloatTensor([[board.get_representation(color)]])
if torch.cuda.is_available():
sample = sample.cuda(0)
sample = Variable(sample)
return self.model(sample)
def get_move(self):
# return self.get_move_alpha_beta()
moves = self.current_board.get_valid_moves(self.color)
# predict value for each possible move
predictions = [(self.__predict_move__(move), move) for move in moves]
# print "Chose move with prediction [%s]" % max(predictions)[0]
self.apply_move(max(predictions)[1])
return self.current_board
def get_move_alpha_beta(self):
move = self.ai.move_search(self.current_board, self.time_limit, self.color, (self.color % 2) + 1)
self.apply_move(move)
return self.current_board
def __predict_move__(self, move):
board = deepcopy(self.current_board)
board.apply_move(move, self.color)
sample = FloatTensor([[board.get_representation(self.color)]])
if torch.cuda.is_available():
sample = sample.cuda(0)
sample = Variable(sample)
return self.model(sample)
