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

23-07-27

Browse Source 
순간 가중치 변화량 조정 0.6 > 0.75
+ evaluate multiproessing false
This commit is contained in:
jung-geun
2023-07-27 14:45:42 +09:00
parent ab937ac71c
commit 64a16ed61a
3 changed files with 56 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
mnist.py
View File

@@ -68,7 +68,7 @@ def make_model():
def random_state(): def random_state():
with open( with open(
"result/mnist/20230720-192726/mean_squared_error_[0.4970000088214874, 0.10073449462652206].json", "result/mnist/20230723-061626/mean_squared_error_[0.6384999752044678, 0.0723000094294548].json",
"r", "r",
) as f: ) as f:
json_ = json.load(f) json_ = json.load(f)
@@ -101,7 +101,7 @@ loss = [
"mean_absolute_percentage_error", "mean_absolute_percentage_error",
] ]
# rs = random_state() rs = random_state()
pso_mnist = Optimizer( pso_mnist = Optimizer(
model, model,
@@ -110,17 +110,18 @@ pso_mnist = Optimizer(
c0=0.3, c0=0.3,
c1=0.5, c1=0.5,
w_min=0.4, w_min=0.4,
w_max=0.9, w_max=0.7,
negative_swarm=0.1, negative_swarm=0.1,
mutation_swarm=0.3, mutation_swarm=0.3,
particle_min=-4, particle_min=-4,
particle_max=4, particle_max=4,
random_state=rs,
) )
best_score = pso_mnist.fit( best_score = pso_mnist.fit(
x_train, x_train,
y_train, y_train,
epochs=200, epochs=250,
save_info=True, save_info=True,
log=2, log=2,
log_name="mnist", log_name="mnist",

67
pso/optimizer.py
View File

@@ -88,39 +88,50 @@ class Optimizer:
self.save_path = None # 저장 위치 self.save_path = None # 저장 위치
self.renewal = "acc" self.renewal = "acc"
self.Dispersion = False self.dispersion = False
self.day = datetime.now().strftime("%Y%m%d-%H%M%S") self.day = datetime.now().strftime("%Y%m%d-%H%M%S")
self.empirical_balance = False self.empirical_balance = False
negative_count = 0 negative_count = 0
print(f"start running time : {self.day}") self.train_summary_writer = [None] * self.n_particles
for i in tqdm(range(self.n_particles), desc="Initializing Particles"): try:
model_ = keras.models.model_from_json(model.to_json()) print(f"start running time : {self.day}")
w_, sh_, len_ = self._encode(model_.get_weights()) for i in tqdm(range(self.n_particles), desc="Initializing Particles"):
w_ = np.random.uniform(particle_min, particle_max, len(w_)) model_ = keras.models.model_from_json(model.to_json())
model_.set_weights(self._decode(w_, sh_, len_)) w_, sh_, len_ = self._encode(model_.get_weights())
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"])
self.particles[i] = Particle(
model_,
loss,
negative=True if i < negative_swarm * self.n_particles else False,
mutation=mutation_swarm,
)
if i < negative_swarm * self.n_particles:
negative_count += 1
# del m, init_weights, w_, sh_, len_
gc.collect()
tf.keras.backend.reset_uids()
tf.keras.backend.clear_session()
print(f"negative swarm : {negative_count} / {self.n_particles}")
print(f"mutation swarm : {mutation_swarm * 100}%")
model_.compile(loss=self.loss, optimizer="sgd", metrics=["accuracy"])
self.particles[i] = Particle(
model_,
loss,
negative=True if i < negative_swarm * self.n_particles else False,
mutation=mutation_swarm,
)
if i < negative_swarm * self.n_particles:
negative_count += 1
# del m, init_weights, w_, sh_, len_
gc.collect() gc.collect()
tf.keras.backend.reset_uids() tf.keras.backend.reset_uids()
tf.keras.backend.clear_session() tf.keras.backend.clear_session()
except KeyboardInterrupt:
print(f"negative swarm : {negative_count} / {self.n_particles}") print("Ctrl + C : Stop Training")
print(f"mutation swarm : {mutation_swarm * 100}%") sys.exit(0)
except MemoryError:
gc.collect() print("Memory Error : Stop Training")
tf.keras.backend.reset_uids() sys.exit(1)
tf.keras.backend.clear_session() except Exception as e:
print(e)
sys.exit(1)
def __del__(self): def __del__(self):
del self.model del self.model
@@ -239,7 +250,7 @@ class Optimizer:
""" """
self.save_path = save_path self.save_path = save_path
self.empirical_balance = empirical_balance self.empirical_balance = empirical_balance
self.Dispersion = dispersion self.dispersion = dispersion
self.renewal = renewal self.renewal = renewal
try: try:
@@ -248,7 +259,6 @@ class Optimizer:
assert log_name is not None, "log_name is None" assert log_name is not None, "log_name is None"
train_log_dir = f"logs/{log_name}/{self.day}/train" train_log_dir = f"logs/{log_name}/{self.day}/train"
self.train_summary_writer = [None] * self.n_particles
for i in range(self.n_particles): for i in range(self.n_particles):
self.train_summary_writer[i] = tf.summary.create_file_writer( self.train_summary_writer[i] = tf.summary.create_file_writer(
train_log_dir + f"/{i}" train_log_dir + f"/{i}"
@@ -263,7 +273,8 @@ class Optimizer:
os.makedirs(save_path, exist_ok=True) os.makedirs(save_path, exist_ok=True)
except ValueError as e: except ValueError as e:
print(e) print(e)
sys.exit(1) except Exception as e:
print(e)
for i in tqdm(range(self.n_particles), desc="Initializing velocity"): for i in tqdm(range(self.n_particles), desc="Initializing velocity"):
p = self.particles[i] p = self.particles[i]
@@ -534,7 +545,7 @@ class Optimizer:
"w_max": self.w_max, "w_max": self.w_max,
"loss_method": self.loss, "loss_method": self.loss,
"empirical_balance": self.empirical_balance, "empirical_balance": self.empirical_balance,
"Dispersion": self.Dispersion, "dispersion": self.dispersion,
"negative_swarm": self.negative_swarm, "negative_swarm": self.negative_swarm,
"mutation_swarm": self.mutation_swarm, "mutation_swarm": self.mutation_swarm,
"random_state_0": self.random_state[0], "random_state_0": self.random_state[0],

25
pso/particle.py
View File

@@ -1,9 +1,6 @@
import gc
import numpy as np import numpy as np
import tensorflow as tf
from tensorflow import keras
from tensorflow import keras
class Particle: class Particle:
""" """
@@ -28,9 +25,9 @@ class Particle:
self.model = model self.model = model
self.loss = loss self.loss = loss
init_weights = self.model.get_weights() init_weights = self.model.get_weights()
i_w_, s_, l_ = self._encode(init_weights) i_w_, i_s, i_l = self._encode(init_weights)
i_w_ = np.random.uniform(-0.5, 0.5, len(i_w_)) i_w_ = np.random.uniform(-0.5, 0.5, len(i_w_))
self.velocities = self._decode(i_w_, s_, l_) self.velocities = self._decode(i_w_, i_s, i_l)
self.negative = negative self.negative = negative
self.mutation = mutation self.mutation = mutation
self.best_score = 0 self.best_score = 0
@@ -38,7 +35,7 @@ class Particle:
self.before_best = init_weights self.before_best = init_weights
self.before_w = 0 self.before_w = 0
del i_w_, s_, l_ del i_w_, i_s, i_l
del init_weights del init_weights
def __del__(self): def __del__(self):
@@ -65,9 +62,9 @@ class Particle:
shape = [] shape = []
for layer in weights: for layer in weights:
shape.append(layer.shape) shape.append(layer.shape)
w_ = layer.reshape(-1) w_tmp = layer.reshape(-1)
length.append(len(w_)) length.append(len(w_tmp))
w_gpu = np.append(w_gpu, w_) w_gpu = np.append(w_gpu, w_tmp)
return w_gpu, shape, length return w_gpu, shape, length
@@ -109,7 +106,7 @@ class Particle:
Returns: Returns:
(float): 점수 (float): 점수
""" """
score = self.model.evaluate(x, y, verbose=0, use_multiprocessing=True) score = self.model.evaluate(x, y, verbose=0)
if renewal == "acc": if renewal == "acc":
if score[1] > self.best_score: if score[1] > self.best_score:
self.best_score = score[1] self.best_score = score[1]
@@ -142,10 +139,10 @@ class Particle:
encode_before, before_sh, before_len = self._encode(weights=self.before_best) encode_before, before_sh, before_len = self._encode(weights=self.before_best)
if (encode_before != encode_g).all(): if (encode_before != encode_g).all():
self.before_w = w self.before_w = w * 0.6
w = w + (self.before_w) w = w + self.before_w
else: else:
self.before_w *= 0.6 self.before_w *= 0.75
w = w + self.before_w w = w + self.before_w
if self.negative: if self.negative: