def test_should_form_match_set(self, cfg):
# given
population = ClassifiersList(cfg=cfg)
situation = Perception('11110000')
# C1 - general condition
c1 = Classifier(cfg=cfg)
# C2 - matching condition
c2 = Classifier(condition='1###0###', cfg=cfg)
# C3 - non-matching condition
c3 = Classifier(condition='0###1###', cfg=cfg)
population.append(c1)
population.append(c2)
population.append(c3)
# when
match_set = ClassifiersList.form_match_set(population,
situation,
cfg=cfg)
# then
assert 2 == len(match_set)
assert c1 in match_set
assert c2 in match_set
def test_should_form_match_set(self, cfg):
# given
cl_1 = Classifier(cfg=cfg)
cl_2 = Classifier(condition='1###0###', cfg=cfg)
cl_3 = Classifier(condition='0###1###', cfg=cfg)
population = ClassifiersList(*[cl_1, cl_2, cl_3])
p0 = Perception('11110000')
# when
match_set = ClassifiersList.form_match_set(population, p0)
# then
assert len(match_set) == 2
assert cl_1 in match_set
assert cl_2 in match_set
len([cl for cl in population if cl.is_reliable()])))
if __name__ == '__main__':
# Play some games
for g in range(GAMES):
action_set = ClassifiersList(cfg=cfg)
prev_state, action, reward, done = None, None, None, False
state = board.reset()
moves = 0
while not done:
player = determine_player(moves) # Determine player
match_set = ClassifiersList.form_match_set(population, state, cfg)
if moves > 0:
action_set.apply_alp(prev_state, action, state,
ALL_MOVES + moves, population, match_set)
action_set.apply_reinforcement_learning(
reward, match_set.get_maximum_fitness())
if cfg.do_ga:
action_set.apply_ga(ALL_MOVES + moves, population,
match_set, state)
# Determine best action
action = match_set.choose_action(cfg.epsilon)
action_set = ClassifiersList.form_action_set(
match_set, action, cfg)
