/
_12_DQN_Nets — kopia.py
41 lines (34 loc) · 1.54 KB
/
_12_DQN_Nets — kopia.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
import tensorflow as tf
from tensorflow.keras import layers,models
from tensorflow.keras.optimizers import Adam
class DQNAgent:
def __init__(self,rows,columns,action_size):
self.rows=rows
self.columns=columns
self.state_size=(self.rows,self.columns)
self.action_size=action_size
#self.memory=deque()
self.gamma=0.95 #discount rate
self.epsilon=1.0 #exploration rate
self.epsilon_min=0.01
self.epsilon_decay=0.995
self.learning_rate=0.001
#Net data
self.output_shape=self.columns
self.modelQ=self.__build_model_Q()
self.modelQ.save('model_Q.h5')
self.modelQ.save_weights('model_weights_Q.h5')
#self.entropy=0
def __build_model_Q(self):
#Neural Net for Deep-Q learning Model
inputs = tf.keras.Input(shape=(self.state_size), name='input_states')
x=tf.keras.layers.Flatten()(inputs)
x=layers.LayerNormalization(name='normalization_layer')(x)
x = layers.Dense(24, activation='elu',kernel_initializer='glorot_normal',name='hidden_layer1')(x)
x = layers.Dense(64, activation='elu',kernel_initializer='glorot_normal',name='hidden_layer2')(x)
outputs = layers.Dense(self.output_shape,activation='linear',kernel_initializer='glorot_normal',name='output_layer')(x)
model = tf.keras.Model(inputs=inputs, outputs=outputs, name='Q_values_model')
model.compile(loss='mse',
optimizer=Adam(lr=self.learning_rate))
model.summary()
return model