jung-geun/PSO
1
0
Fork 0
mirror of https://github.com/jung-geun/PSO.git synced 2025-12-20 04:50:45 +09:00
Code Issues Packages Projects Releases Wiki Activity

23-10-18

Browse Source 
batch size 적용 -> 속도 개선
역전파 1회 적용 -> 조기 수렴을 일부 방지
This commit is contained in:
jung-geun
2023-10-18 14:42:59 +09:00
parent f32433d935
commit dfc0df7d48
10 changed files with 157 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

83
pso/optimizer.py
View File

@@ -106,7 +106,11 @@ class Optimizer:
w_ = np.random.uniform(particle_min, particle_max, len(w_))
model_.set_weights(self._decode(w_, sh_, len_))
model_.compile(loss=self.loss, optimizer="sgd", metrics=["accuracy"])
model_.compile(
loss=self.loss,
optimizer="sgd",
metrics=["accuracy"]
)
self.particles[i] = Particle(
model_,
loss,
@@ -224,6 +228,20 @@ class Optimizer:
else:
return 1 + np.abs(score_)
class _batch_generator:
def __init__(self, x, y, batch_size: int = 32):
self.batch_size = batch_size
self.index = 0
dataset = tf.data.Dataset.from_tensor_slices((x, y))
self.dataset = list(dataset.batch(batch_size))
self.max_index = len(dataset) // batch_size
def next(self):
self.index += 1
if self.index >= self.max_index:
self.index = 0
return self.dataset[self.index][0], self.dataset[self.index][1]
def fit(
self,
x,
@@ -237,6 +255,7 @@ class Optimizer:
empirical_balance: bool = False,
dispersion: bool = False,
check_point: int = None,
batch_size: int = 128,
):
"""
# Args:
@@ -250,6 +269,7 @@ class Optimizer:
empirical_balance : bool - True : EBPSO, False : PSO,
dispersion : bool - True : g_best 의 값을 분산시켜 전역해를 찾음, False : g_best 의 값만 사용
check_point : int - 저장할 위치 - None : 저장 안함
batch_size : int - batch size default : 128
"""
self.save_path = save_path
self.empirical_balance = empirical_balance
@@ -278,10 +298,33 @@ class Optimizer:
except ValueError as ve:
sys.exit(ve)
model_ = keras.models.model_from_json(self.model.to_json())
model_.compile(loss=self.loss, optimizer="adam", metrics=["accuracy"])
model_.fit(x, y, epochs=1, batch_size=64, verbose=0)
score = model_.evaluate(x, y, verbose=1)
if renewal == "acc":
self.g_best_score[0] = score[1]
self.g_best_score[1] = score[0]
else:
self.g_best_score[0] = score[0]
self.g_best_score[1] = score[1]
self.g_best = model_.get_weights()
self.g_best_ = model_.get_weights()
del model_
dataset = self._batch_generator(x, y, batch_size=batch_size)
for i in tqdm(range(self.n_particles), desc="Initializing velocity"):
p = self.particles[i]
local_score = p.get_score(x, y, renewal=renewal)
x_batch, y_batch = dataset.next()
local_score = p.get_score(x_batch, y_batch, 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]
@@ -329,7 +372,7 @@ class Optimizer:
epoch_sum = 0
epochs_pbar = tqdm(
range(epochs),
desc=f"best {self.g_best_score[0]:.4f}|{self.g_best_score[1]:.4f}",
desc=f"best {self.g_best_score[0]:.4f} | {self.g_best_score[1]:.4f}",
ascii=True,
leave=True,
position=0,
@@ -355,6 +398,8 @@ class Optimizer:
g_best = self.g_best
x_batch, y_batch = dataset.next()
if dispersion:
ts = self.particle_min + np.random.rand() * (
self.particle_max - self.particle_min
@@ -367,7 +412,9 @@ class Optimizer:
if empirical_balance:
if np.random.rand() < np.exp(-(epoch) / epochs):
w_p_ = self._f(x, y, self.particles[i].get_best_weights())
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_)
@@ -417,8 +464,8 @@ class Optimizer:
del p_
score = self.particles[i].step_w(
x,
y,
x_batch,
y_batch,
self.c0,
self.c1,
w,
@@ -431,13 +478,15 @@ class Optimizer:
else:
score = self.particles[i].step(
x, y, self.c0, self.c1, w, g_best, renewal=renewal
x_batch, y_batch, self.c0, self.c1, w, g_best, renewal=renewal
)
if log == 2:
with self.train_summary_writer[i].as_default():
tf.summary.scalar("loss", score[0], step=epoch + 1)
tf.summary.scalar("accuracy", score[1], step=epoch + 1)
tf.summary.scalar(
"accuracy", score[1], step=epoch + 1
)
if renewal == "acc":
if score[1] >= max_score:
@@ -447,11 +496,13 @@ class Optimizer:
if score[1] >= self.g_best_score[0]:
if score[1] > self.g_best_score[0]:
self.g_best_score[0] = score[1]
self.g_best = self.particles[i].get_best_weights()
self.g_best = self.particles[i].get_best_weights(
)
else:
if score[0] < self.g_best_score[1]:
self.g_best_score[1] = score[0]
self.g_best = self.particles[i].get_best_weights()
self.g_best = self.particles[i].get_best_weights(
)
epochs_pbar.set_description(
f"best {self.g_best_score[0]:.4f} | {self.g_best_score[1]:.4f}"
)
@@ -463,11 +514,13 @@ class Optimizer:
if score[0] <= self.g_best_score[1]:
if score[0] < self.g_best_score[1]:
self.g_best_score[1] = score[0]
self.g_best = self.particles[i].get_best_weights()
self.g_best = self.particles[i].get_best_weights(
)
else:
if score[1] > self.g_best_score[0]:
self.g_best_score[0] = score[1]
self.g_best = self.particles[i].get_best_weights()
self.g_best = self.particles[i].get_best_weights(
)
epochs_pbar.set_description(
f"best {self.g_best_score[0]:.4f} | {self.g_best_score[1]:.4f}"
)
@@ -505,7 +558,8 @@ class Optimizer:
if check_point is not None:
if epoch % check_point == 0:
os.makedirs(f"./{save_path}/{self.day}", exist_ok=True)
self._check_point_save(f"./{save_path}/{self.day}/ckpt-{epoch}")
self._check_point_save(
f"./{save_path}/{self.day}/ckpt-{epoch}")
gc.collect()
tf.keras.backend.reset_uids()
@@ -513,10 +567,13 @@ class Optimizer:
except KeyboardInterrupt:
print("Ctrl + C : Stop Training")
except MemoryError:
print("Memory Error : Stop Training")
except Exception as e:
print(e)
finally:
self.model_save(save_path)
print("model save")