mirror of
https://github.com/jung-geun/PSO.git
synced 2025-12-19 20:44:39 +09:00
jung-geun
@@ -1,9 +1,9 @@
|
||||
from .optimizer import Optimizer
|
||||
from .particle import Particle
|
||||
from .optimizer import Optimizer as optimizer
|
||||
from .particle import Particle as particle
|
||||
|
||||
__version__ = "0.1.6"
|
||||
|
||||
__all__ = [
|
||||
"Optimizer",
|
||||
"Particle",
|
||||
"optimizer",
|
||||
"particle",
|
||||
]
|
||||
|
||||
127
pso/optimizer.py
127
pso/optimizer.py
@@ -15,8 +15,8 @@ gpus = tf.config.experimental.list_physical_devices("GPU")
|
||||
if gpus:
|
||||
try:
|
||||
tf.config.experimental.set_memory_growth(gpus[0], True)
|
||||
except RuntimeError as e:
|
||||
print(e)
|
||||
except RuntimeError as r:
|
||||
print(r)
|
||||
|
||||
|
||||
class Optimizer:
|
||||
@@ -81,10 +81,13 @@ class Optimizer:
|
||||
self.w_max = w_max # 최대 관성 수치
|
||||
self.negative_swarm = negative_swarm # 최적해와 반대로 이동할 파티클 비율 - 0 ~ 1 사이의 값
|
||||
self.mutation_swarm = mutation_swarm # 관성을 추가로 사용할 파티클 비율 - 0 ~ 1 사이의 값
|
||||
self.particle_min = particle_min # 가중치 초기화 최소값
|
||||
self.particle_max = particle_max
|
||||
self.g_best_score = [0, np.inf] # 최고 점수 - 시작은 0으로 초기화
|
||||
self.g_best = None # 최고 점수를 받은 가중치
|
||||
self.g_best_ = None # 최고 점수를 받은 가중치 - 값의 분산을 위한 변수
|
||||
self.avg_score = 0 # 평균 점수
|
||||
self.sigma = 1.0
|
||||
|
||||
self.save_path = None # 저장 위치
|
||||
self.renewal = "acc"
|
||||
@@ -124,14 +127,11 @@ class Optimizer:
|
||||
tf.keras.backend.reset_uids()
|
||||
tf.keras.backend.clear_session()
|
||||
except KeyboardInterrupt:
|
||||
print("Ctrl + C : Stop Training")
|
||||
sys.exit(0)
|
||||
sys.exit("Ctrl + C : Stop Training")
|
||||
except MemoryError:
|
||||
print("Memory Error : Stop Training")
|
||||
sys.exit(1)
|
||||
sys.exit("Memory Error : Stop Training")
|
||||
except Exception as e:
|
||||
print(e)
|
||||
sys.exit(1)
|
||||
sys.exit(e)
|
||||
|
||||
def __del__(self):
|
||||
del self.model
|
||||
@@ -147,6 +147,7 @@ class Optimizer:
|
||||
del self.g_best
|
||||
del self.g_best_
|
||||
del self.avg_score
|
||||
|
||||
gc.collect()
|
||||
tf.keras.backend.reset_uids()
|
||||
tf.keras.backend.clear_session()
|
||||
@@ -167,9 +168,9 @@ class Optimizer:
|
||||
shape = []
|
||||
for layer in weights:
|
||||
shape.append(layer.shape)
|
||||
w_ = layer.reshape(-1)
|
||||
length.append(len(w_))
|
||||
w_gpu = np.append(w_gpu, w_)
|
||||
w_tmp = layer.reshape(-1)
|
||||
length.append(len(w_tmp))
|
||||
w_gpu = np.append(w_gpu, w_tmp)
|
||||
|
||||
del weights
|
||||
|
||||
@@ -191,18 +192,17 @@ class Optimizer:
|
||||
start = 0
|
||||
for i in range(len(shape)):
|
||||
end = start + length[i]
|
||||
w_ = weight[start:end]
|
||||
w_ = np.reshape(w_, shape[i])
|
||||
weights.append(w_)
|
||||
w_tmp = weight[start:end]
|
||||
w_tmp = np.reshape(w_tmp, shape[i])
|
||||
weights.append(w_tmp)
|
||||
start = end
|
||||
|
||||
del weight
|
||||
del shape
|
||||
del length
|
||||
del weight, shape, length
|
||||
del start, end, w_tmp
|
||||
|
||||
return weights
|
||||
|
||||
def f(self, x, y, weights):
|
||||
def _f(self, x, y, weights):
|
||||
"""
|
||||
EBPSO의 목적함수 (예상)
|
||||
|
||||
@@ -215,11 +215,16 @@ class Optimizer:
|
||||
(float): 목적 함수 값
|
||||
"""
|
||||
self.model.set_weights(weights)
|
||||
score = self.model.evaluate(x, y, verbose=0)[1]
|
||||
if score > 0:
|
||||
return 1 / (1 + score)
|
||||
score = self.model.evaluate(x, y, verbose=0)
|
||||
if self.renewal == "acc":
|
||||
score_ = score[1]
|
||||
else:
|
||||
return 1 + np.abs(score)
|
||||
score_ = score[0]
|
||||
|
||||
if score_ > 0:
|
||||
return 1 / (1 + score_)
|
||||
else:
|
||||
return 1 + np.abs(score_)
|
||||
|
||||
def fit(
|
||||
self,
|
||||
@@ -253,6 +258,7 @@ class Optimizer:
|
||||
self.dispersion = dispersion
|
||||
|
||||
self.renewal = renewal
|
||||
particle_sum = 0 # x_j
|
||||
try:
|
||||
train_log_dir = "logs/fit/" + self.day
|
||||
if log == 2:
|
||||
@@ -269,17 +275,17 @@ class Optimizer:
|
||||
raise ValueError("save_path is None")
|
||||
else:
|
||||
self.save_path = save_path
|
||||
if not os.path.exists(save_path):
|
||||
os.makedirs(save_path, exist_ok=True)
|
||||
if not os.path.exists(f"{save_path}/{self.day}"):
|
||||
os.makedirs(f"{save_path}/{self.day}", exist_ok=True)
|
||||
except ValueError as e:
|
||||
print(e)
|
||||
sys.exit(e)
|
||||
except Exception as e:
|
||||
print(e)
|
||||
sys.exit(e)
|
||||
|
||||
for i in tqdm(range(self.n_particles), desc="Initializing velocity"):
|
||||
p = self.particles[i]
|
||||
local_score = p.get_score(x, y, renewal=renewal)
|
||||
|
||||
particle_sum += local_score[1]
|
||||
if renewal == "acc":
|
||||
if local_score[1] > self.g_best_score[0]:
|
||||
self.g_best_score[0] = local_score[1]
|
||||
@@ -301,7 +307,7 @@ class Optimizer:
|
||||
|
||||
if log == 1:
|
||||
with open(
|
||||
f"./{save_path}/{self.day}_{self.n_particles}_{epochs}_{self.c0}_{self.c1}_{self.w_min}_{renewal}.csv",
|
||||
f"./{save_path}/{self.day}/{self.n_particles}_{epochs}_{self.c0}_{self.c1}_{self.w_min}_{renewal}.csv",
|
||||
"a",
|
||||
) as f:
|
||||
f.write(f"{local_score[0]}, {local_score[1]}")
|
||||
@@ -316,14 +322,15 @@ class Optimizer:
|
||||
tf.summary.scalar("accuracy", local_score[1], step=0)
|
||||
|
||||
del local_score
|
||||
gc.collect()
|
||||
tf.keras.backend.reset_uids()
|
||||
tf.keras.backend.clear_session()
|
||||
# gc.collect()
|
||||
# tf.keras.backend.reset_uids()
|
||||
# tf.keras.backend.clear_session()
|
||||
print(
|
||||
f"initial g_best_score : {self.g_best_score[0] if self.renewal == 'acc' else self.g_best_score[1]}"
|
||||
)
|
||||
|
||||
try:
|
||||
epoch_sum = 0
|
||||
epochs_pbar = tqdm(
|
||||
range(epochs),
|
||||
desc=f"best {self.g_best_score[0]:.4f}|{self.g_best_score[1]:.4f}",
|
||||
@@ -332,8 +339,11 @@ class Optimizer:
|
||||
position=0,
|
||||
)
|
||||
for epoch in epochs_pbar:
|
||||
particle_avg = particle_sum / self.n_particles # x_j
|
||||
particle_sum = 0
|
||||
max_score = 0
|
||||
min_loss = np.inf
|
||||
# epoch_particle_sum = 0
|
||||
part_pbar = tqdm(
|
||||
range(len(self.particles)),
|
||||
desc=f"acc : {max_score:.4f} loss : {min_loss:.4f}",
|
||||
@@ -347,20 +357,22 @@ class Optimizer:
|
||||
f"acc : {max_score:.4f} loss : {min_loss:.4f}"
|
||||
)
|
||||
|
||||
g_best = self.g_best
|
||||
|
||||
if dispersion:
|
||||
ts = self.c0 + np.random.rand() * (self.c1 - self.c0)
|
||||
ts = self.particle_min + np.random.rand() * (
|
||||
self.particle_max - self.particle_min
|
||||
)
|
||||
g_, g_sh, g_len = self._encode(self.g_best)
|
||||
decrement = (epochs - (epoch) + 1) / epochs
|
||||
decrement = (epochs - epoch + 1) / epochs
|
||||
g_ = (1 - decrement) * g_ + decrement * ts
|
||||
self.g_best_ = self._decode(g_, g_sh, g_len)
|
||||
g_best = self.g_best_
|
||||
else:
|
||||
g_best = self.g_best
|
||||
|
||||
if empirical_balance:
|
||||
if np.random.rand() < np.exp(-(epoch) / epochs):
|
||||
w_p_ = self.f(x, y, self.particles[i].get_best_weights())
|
||||
w_g_ = self.f(x, y, self.g_best)
|
||||
w_p_ = self._f(x, y, self.particles[i].get_best_weights())
|
||||
w_g_ = self._f(x, y, self.g_best)
|
||||
w_p = w_p_ / (w_p_ + w_g_)
|
||||
w_g = w_p_ / (w_p_ + w_g_)
|
||||
|
||||
@@ -371,13 +383,35 @@ class Optimizer:
|
||||
p_b = self.particles[i].get_best_score()
|
||||
g_a = self.avg_score
|
||||
l_b = p_b - g_a
|
||||
l_b = np.sqrt(np.power(l_b, 2))
|
||||
p_ = (
|
||||
sigma_post = np.sqrt(np.power(l_b, 2))
|
||||
sigma_pre = (
|
||||
1
|
||||
/ (self.n_particles * np.linalg.norm(self.c1 - self.c0))
|
||||
* l_b
|
||||
/ (
|
||||
self.n_particles
|
||||
* np.linalg.norm(
|
||||
self.particle_max - self.particle_min
|
||||
)
|
||||
)
|
||||
* sigma_post
|
||||
)
|
||||
p_ = np.exp(-1 * p_)
|
||||
p_ = np.exp(-1 * sigma_pre * sigma_post)
|
||||
|
||||
# p_ = (
|
||||
# 1
|
||||
# / (self.n_particles * np.linalg.norm(self.particle_max - self.particle_min))
|
||||
# * np.exp(
|
||||
# -np.power(l_b, 2) / (2 * np.power(self.sigma, 2))
|
||||
# )
|
||||
# )
|
||||
# g_ = (
|
||||
# 1
|
||||
# / np.linalg.norm(self.c1 - self.c0)
|
||||
# * np.exp(
|
||||
# -np.power(l_b, 2) / (2 * np.power(self.sigma, 2))
|
||||
# )
|
||||
# )
|
||||
# w_p = p_ / (p_ + g_)
|
||||
# w_g = g_ / (p_ + g_)
|
||||
w_p = p_
|
||||
w_g = 1 - p_
|
||||
|
||||
@@ -397,6 +431,7 @@ class Optimizer:
|
||||
w_g,
|
||||
renewal=renewal,
|
||||
)
|
||||
epoch_sum += np.power(score[1] - particle_avg, 2)
|
||||
|
||||
else:
|
||||
score = self.particles[i].step(
|
||||
@@ -457,10 +492,11 @@ class Optimizer:
|
||||
epochs_pbar.set_description(
|
||||
f"best {self.g_best_score[0]:.4f} | {self.g_best_score[1]:.4f}"
|
||||
)
|
||||
particle_sum += score[1]
|
||||
|
||||
if log == 1:
|
||||
with open(
|
||||
f"./{save_path}/{self.day}_{self.n_particles}_{epochs}_{self.c0}_{self.c1}_{self.w_min}_{renewal}.csv",
|
||||
f"./{save_path}/{self.day}/{self.n_particles}_{epochs}_{self.c0}_{self.c1}_{self.w_min}_{renewal}.csv",
|
||||
"a",
|
||||
) as f:
|
||||
f.write(f"{score[0]}, {score[1]}")
|
||||
@@ -468,9 +504,6 @@ class Optimizer:
|
||||
f.write(", ")
|
||||
else:
|
||||
f.write("\n")
|
||||
# gc.collect()
|
||||
# tf.keras.backend.reset_uids()
|
||||
# tf.keras.backend.clear_session()
|
||||
part_pbar.refresh()
|
||||
|
||||
if check_point is not None:
|
||||
@@ -536,7 +569,7 @@ class Optimizer:
|
||||
path (str, optional): 저장 위치. Defaults to "./result".
|
||||
"""
|
||||
json_save = {
|
||||
"name": f"{self.day}_{self.n_particles}_{self.c0}_{self.c1}_{self.w_min}.h5",
|
||||
"name": f"{self.day}/{self.n_particles}_{self.c0}_{self.c1}_{self.w_min}.h5",
|
||||
"n_particles": self.n_particles,
|
||||
"score": self.g_best_score,
|
||||
"c0": self.c0,
|
||||
|
||||
@@ -2,6 +2,7 @@ import numpy as np
|
||||
|
||||
from tensorflow import keras
|
||||
|
||||
|
||||
class Particle:
|
||||
"""
|
||||
Particle Swarm Optimization의 Particle을 구현한 클래스
|
||||
@@ -112,8 +113,6 @@ class Particle:
|
||||
self.best_score = score[1]
|
||||
self.best_weights = self.model.get_weights()
|
||||
elif renewal == "loss":
|
||||
if score[0] == "nan":
|
||||
score[0] = np.inf
|
||||
if score[0] < self.best_score:
|
||||
self.best_score = score[0]
|
||||
self.best_weights = self.model.get_weights()
|
||||
@@ -134,11 +133,11 @@ class Particle:
|
||||
encode_v, v_sh, v_len = self._encode(weights=self.velocities)
|
||||
encode_p, p_sh, p_len = self._encode(weights=self.best_weights)
|
||||
encode_g, g_sh, g_len = self._encode(weights=g_best)
|
||||
r0 = np.random.rand()
|
||||
r1 = np.random.rand()
|
||||
encode_before, before_sh, before_len = self._encode(weights=self.before_best)
|
||||
r_0 = np.random.rand()
|
||||
r_1 = np.random.rand()
|
||||
|
||||
if (encode_before != encode_g).all():
|
||||
if not np.array_equal(encode_before, encode_g, equal_nan=True):
|
||||
self.before_w = w * 0.6
|
||||
w = w + self.before_w
|
||||
else:
|
||||
@@ -148,14 +147,14 @@ class Particle:
|
||||
if self.negative:
|
||||
new_v = (
|
||||
w * encode_v
|
||||
+ -1 * local_rate * r0 * (encode_p - encode_w)
|
||||
+ -1 * global_rate * r1 * (encode_g - encode_w)
|
||||
+ local_rate * r_0 * (encode_p - encode_w)
|
||||
+ -1 * global_rate * r_1 * (encode_g - encode_w)
|
||||
)
|
||||
else:
|
||||
new_v = (
|
||||
w * encode_v
|
||||
+ local_rate * r0 * (encode_p - encode_w)
|
||||
+ global_rate * r1 * (encode_g - encode_w)
|
||||
+ local_rate * r_0 * (encode_p - encode_w)
|
||||
+ global_rate * r_1 * (encode_g - encode_w)
|
||||
)
|
||||
|
||||
if np.random.rand() < self.mutation:
|
||||
@@ -169,7 +168,7 @@ class Particle:
|
||||
del encode_p, p_sh, p_len
|
||||
del encode_g, g_sh, g_len
|
||||
del encode_before, before_sh, before_len
|
||||
del r0, r1
|
||||
del r_0, r_1
|
||||
|
||||
def _update_velocity_w(self, local_rate, global_rate, w, w_p, w_g, g_best):
|
||||
"""
|
||||
@@ -188,20 +187,27 @@ class Particle:
|
||||
encode_v, v_sh, v_len = self._encode(weights=self.velocities)
|
||||
encode_p, p_sh, p_len = self._encode(weights=self.best_weights)
|
||||
encode_g, g_sh, g_len = self._encode(weights=g_best)
|
||||
r0 = np.random.rand()
|
||||
r1 = np.random.rand()
|
||||
encode_before, before_sh, before_len = self._encode(weights=self.before_best)
|
||||
r_0 = np.random.rand()
|
||||
r_1 = np.random.rand()
|
||||
|
||||
if not np.array_equal(encode_before, encode_g, equal_nan=True):
|
||||
self.before_w = w * 0.6
|
||||
w = w + self.before_w
|
||||
else:
|
||||
self.before_w *= 0.75
|
||||
w = w + self.before_w
|
||||
if self.negative:
|
||||
new_v = (
|
||||
w * encode_v
|
||||
+ -1 * local_rate * r0 * (w_p * encode_p - encode_w)
|
||||
+ -1 * global_rate * r1 * (w_g * encode_g - encode_w)
|
||||
+ local_rate * r_0 * (w_p * encode_p - encode_w)
|
||||
+ -1 * global_rate * r_1 * (w_g * encode_g - encode_w)
|
||||
)
|
||||
else:
|
||||
new_v = (
|
||||
w * encode_v
|
||||
+ local_rate * r0 * (w_p * encode_p - encode_w)
|
||||
+ global_rate * r1 * (w_g * encode_g - encode_w)
|
||||
+ local_rate * r_0 * (w_p * encode_p - encode_w)
|
||||
+ global_rate * r_1 * (w_g * encode_g - encode_w)
|
||||
)
|
||||
|
||||
if np.random.rand() < self.mutation:
|
||||
@@ -214,7 +220,8 @@ class Particle:
|
||||
del encode_v, v_sh, v_len
|
||||
del encode_p, p_sh, p_len
|
||||
del encode_g, g_sh, g_len
|
||||
del r0, r1
|
||||
del encode_before, before_sh, before_len
|
||||
del r_0, r_1
|
||||
|
||||
def _update_weights(self):
|
||||
"""
|
||||
|
||||
Reference in New Issue
Block a user