'''
Created on 2018-4-4

'''
keras.layers.core.Dense(
units, #代表該層的輸出維度
activation=None, #激活函數(shù).但是默認(rèn) liner
use_bias=True, #是否使用b
kernel_initializer='glorot_uniform', #初始化w權(quán)重，keras/initializers.py
bias_initializer='zeros', #初始化b權(quán)重
kernel_regularizer=None, #施加在權(quán)重w上的正則項(xiàng),keras/regularizer.py
bias_regularizer=None, #施加在偏置向量b上的正則項(xiàng)
activity_regularizer=None, #施加在輸出上的正則項(xiàng)
kernel_constraint=None, #施加在權(quán)重w上的約束項(xiàng)
bias_constraint=None #施加在偏置b上的約束項(xiàng)
)

# 所實(shí)現(xiàn)的運(yùn)算是output = activation(dot(input, kernel)+bias)
# model.add(Dense(units=64, activation='relu', input_dim=784))

# keras初始化所有激活函數(shù),activation:
# keras\activations.py
# keras\backend\cntk_backend.py
# import cntk as C
# 1.softmax：
#       對(duì)輸入數(shù)據(jù)的最后一維進(jìn)行softmax，一般用在輸出層;
#   ndim == 2,K.softmax(x),其實(shí)調(diào)用的是cntk，是一個(gè)模塊;
#   ndim >= 2,e = K.exp(x - K.max(x)),s = K.sum(e),return e / s
# 2.elu
#   K.elu(x)
# 3.selu: 可伸縮的指數(shù)線性單元
#   alpha = 1.6732632423543772848170429916717
#   scale = 1.0507009873554804934193349852946
#   return scale * K.elu(x, alpha)
# 4.softplus
#   C.softplus(x)
# 5.softsign
#   return x / (1 + C.abs(x))
# 6.relu
#   def relu(x, alpha=0., max_value=None):
#     if alpha != 0.:
#       negative_part = C.relu(-x)
#     x = C.relu(x)
#     if max_value is not None:
#       x = C.clip(x, 0.0, max_value)
#     if alpha != 0.:
#       x -= alpha * negative_part
#     return x
# 7.tanh
#   return C.tanh(x)
# 8.sigmoid
#   return C.sigmoid(x)
# 9.hard_sigmoid
#   x = (0.2 * x) + 0.5
#   x = C.clip(x, 0.0, 1.0)
#   return x
# 10.linear
#   return x

# keras初始化所有方法，initializer:
# Zeros
# Ones
# Constant(固定一個(gè)值)
# RandomNormal(正態(tài)分布)
# RandomUniform(均勻分布)
# TruncatedNormal(截尾高斯分布,神經(jīng)網(wǎng)絡(luò)權(quán)重和濾波器的推薦初始化方法)
# VarianceScaling(該初始化方法能夠自適應(yīng)目標(biāo)張量的shape)
# Orthogonal(隨機(jī)正交矩陣初始化)
# Identiy(單位矩陣初始化，僅適用于2D方陣)
# lecun_uniform(LeCun均勻分布初始化)
# lecun_normal(LeCun正態(tài)分布初始化)
# glorot_normal(Glorot正態(tài)分布初始化)
# glorot_uniform(Glorot均勻分布初始化)
# he_normal(He正態(tài)分布初始化)
# he_uniform(He均勻分布初始化,Keras中文文檔寫錯(cuò)了)

# keras正則化，regularizer:
# import backend as K
# L1: regularization += K.sum(self.l1 * K.abs(x))
# L2: regularization += K.sum(self.l2 * K.square(x))

keras.layers.Dense(units, 
  activation=None, 
  use_bias=True, 
  kernel_initializer='glorot_uniform', 
  bias_initializer='zeros', 
  kernel_regularizer=None, 
  bias_regularizer=None, 
   activity_regularizer=None, 
  kernel_constraint=None, 
  bias_constraint=None)

from keras.layers import Dense

# 作為 Sequential 模型的第一層
model = Sequential()
model.add(Dense(32, input_shape=(16,)))
# 現(xiàn)在模型就會(huì)以尺寸為 (*, 16) 的數(shù)組作為輸入，
# 其輸出數(shù)組的尺寸為 (*, 32)

# 在第一層之后，你就不再需要指定輸入的尺寸了：
model.add(Dense(32))

from keras.layers import Dense

model = Sequential()
model.add(Dense(512, activation= 'sigmoid', input_dim= 2, use_bias= True))