滑动平均
滑动平均(影子值):记录了每个参数一段时间内过往值的平均,增加了模型的泛化性。
针对所有参数:w和b
(像是给参数加了影子,参数变化,影子缓慢追随)
影子 = 衰减率影子+(1-衰减率)参数 影子初值 = 参数初值
衰减率 = min{MOVING_AVERAGE_DECAY,}
ema = tf.train.ExponentialMovingAverage(衰减率MOVING_AVERAGE_DECAY(是一个超参数),当前轮数global_step)
ema_op = ema.apply([])
ema_op = ema.apply(tf.trainable_variables())
每运行此句,所有待优化的参数求滑动平均
with tf.control_dependencies([train_step,ema_op]):
train_op = tf.no_op(name = "train")
ema.average(参数名) 查看某参数的滑动平均值
import tensorflow as tf
#1. 定义变量及滑动平均类
#定义一个32位浮点变量,初始值为0.0 这个代码就是不断更新w1参数,优化w1参数,滑动平均做了个w1的影子
w1 = tf.Variable(0, dtype=tf.float32)
#定义num_updates(NN的迭代轮数),初始值为0,不可被优化(训练),这个参数不训练
global_step = tf.Variable(0, trainable=False)
#实例化滑动平均类,给衰减率为0.99,当前轮数global_step
MOVING_AVERAGE_DECAY = 0.99
ema = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)
#ema.apply后的括号里是更新列表,每次运行sess.run(ema_op)时,对更新列表中的元素求滑动平均值。
#在实际应用中会使用tf.trainable_variables()自动将所有待训练的参数汇总为列表
#ema_op = ema.apply([w1])
ema_op = ema.apply(tf.trainable_variables()) #滑动平均节点
#2. 查看不同迭代中变量取值的变化。
with tf.Session() as sess:
# 初始化
init_op = tf.global_variables_initializer()
sess.run(init_op)
#用ema.average(w1)获取w1滑动平均值 (要运行多个节点,作为列表中的元素列出,写在sess.run中)
#打印出当前参数w1和w1滑动平均值
print ("current global_step:", sess.run(global_step))
print ("current w1", sess.run([w1, ema.average(w1)]) )
# 参数w1的值赋为1
sess.run(tf.assign(w1, 1))
sess.run(ema_op)
print ("current global_step:", sess.run(global_step))
print ("current w1", sess.run([w1, ema.average(w1)]) )
# 更新global_step和w1的值,模拟出轮数为100时,参数w1变为10, 以下代码global_step保持为100,每次执行滑动平均操作,影子值会更新
sess.run(tf.assign(global_step, 100))
sess.run(tf.assign(w1, 10))
sess.run(ema_op)
print ("current global_step:", sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
# 每次sess.run会更新一次w1的滑动平均值
sess.run(ema_op)
print ("current global_step:" , sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
sess.run(ema_op)
print ("current global_step:" , sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
sess.run(ema_op)
print ("current global_step:" , sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
sess.run(ema_op)
print ("current global_step:" , sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
sess.run(ema_op)
print ("current global_step:" , sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
sess.run(ema_op)
print ("current global_step:" , sess.run(global_step))
print ("current w1:", sess.run([w1, ema.average(w1)]))
#随后每执行一次,参数w1的滑动平均都向w1逼近
#更改MOVING_AVERAGE_DECAY 为 0.1 看影子追随速度
#衰减率减小后,影子追随速度增加
正则化缓解过拟合
正则化在损失函数中引入模型复杂度指标,利用给W加权值,弱化了训练数据的噪声(一般不正则化b)
loss(w) = tf.contirb.layers.l1_regularizer(REGULARIZER)(w)
loss(w) = tf.contirb.layers.l2_regularizer(REGULARIZER)(w)
tf.add_to_collection("losses",tf.contrib.layers.l2_regularizer(regularizer)(w))
loss = cem+tf.add_n(tf.get_collection("losses"))
BATCH_SIZE = 30
seed = 2
#基于seed产生随机数
rdm = np.random.RandomState(seed)
print("rdm:\n",rdm)
#随机数返回300行2列的矩阵,表示300组坐标点(x0,x1)作为输入数据集
X = rdm.randn(300,2)
print("X:\n",X)
#从X这个300行2列的矩阵中取出一行,判断如果两个坐标的平方和小于2,给Y赋值1,其余赋值0
#作为输入数据集的标签(正确答案)
Y_ = [int(x0*x0 + x1*x1 <2) for (x0,x1) in X]
print("Y_:\n",Y_)
#遍历Y中的每个元素,1赋值'red'其余赋值'blue',这样可视化显示时人可以直观区分
Y_c = [['red' if y else 'blue'] for y in Y_]
#对数据集X和标签Y进行shape整理,第一个元素为-1表示,随第二个参数计算得到,第二个元素表示多少列,把X整理为n行2列,把Y整理为n行1列
X = np.vstack(X).reshape(-1,2)
Y_ = np.vstack(Y_).reshape(-1,1)
print("Y_c:\n",Y_c)
print("X:\n",X)
print("Y_:\n",Y_)
np.squeeze(Y_c) #从数组的形状中删除单维度条目,即把shape中为1的维度去掉
#用plt.scatter画出数据集X各行中第0列元素和第1列元素的点即各行的(x0,x1),用各行Y_c对应的值表示颜色(c是color的缩写)
plt.scatter(X[:,0], X[:,1], c=np.squeeze(Y_c))
plt.show()
![图片.png](https://img.haomeiwen.com/i11956727/0834a86d206bd429.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
#定义神经网络的输入、参数和输出,定义前向传播过程
def get_weight(shape, regularizer): #w的shape,w的正则化权重
w = tf.Variable(tf.random_normal(shape), dtype=tf.float32)
tf.add_to_collection('losses', tf.contrib.layers.l2_regularizer(regularizer)(w))
return w
def get_bias(shape):
b = tf.Variable(tf.constant(0.01, shape=shape))
return b
x = tf.placeholder(tf.float32, shape=(None, 2))
y_ = tf.placeholder(tf.float32, shape=(None, 1))
w1 = get_weight([2,11], 0.01)
b1 = get_bias([11])
y1 = tf.nn.relu(tf.matmul(x, w1)+b1)
w2 = get_weight([11,1], 0.01)
b2 = get_bias([1])
y = tf.matmul(y1, w2)+b2
#定义损失函数
loss_mse = tf.reduce_mean(tf.square(y-y_))
loss_total = loss_mse + tf.add_n(tf.get_collection('losses'))
#定义反向传播方法:不含正则化
train_step = tf.train.AdamOptimizer(0.0001).minimize(loss_mse)
with tf.Session() as sess:
init_op = tf.global_variables_initializer()
sess.run(init_op)
STEPS = 40000
for i in range(STEPS):
start = (i*BATCH_SIZE) % 300
end = start + BATCH_SIZE
sess.run(train_step, feed_dict={x:X[start:end], y_:Y_[start:end]})
if i % 2000 == 0:
loss_mse_v = sess.run(loss_mse, feed_dict={x:X, y_:Y_})
print("After %d steps, loss is: %f" %(i, loss_mse_v))
#xx在-3到3之间以步长为0.01,yy在-3到3之间以步长0.01,生成二维网格坐标点
xx, yy = np.mgrid[-3:3:.01, -3:3:.01]
#将xx , yy拉直,并合并成一个2列的矩阵,得到一个网格坐标点的集合
grid = np.c_[xx.ravel(), yy.ravel()]
#将网格坐标点喂入神经网络 ,probs为输出
probs = sess.run(y, feed_dict={x:grid})
#probs的shape调整成xx的样子
probs = probs.reshape(xx.shape)
print ("w1:\n",sess.run(w1))
print ("b1:\n",sess.run(b1))
print ("w2:\n",sess.run(w2) )
print ("b2:\n",sess.run(b2))
plt.scatter(X[:,0], X[:,1], c=np.squeeze(Y_c))
plt.contour(xx, yy, probs, levels=[.5])
plt.show()
![图片.png](https://img.haomeiwen.com/i11956727/b86c2a7ef1e42d17.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
#定义反向传播方法:包含正则化
train_step = tf.train.AdamOptimizer(0.0001).minimize(loss_total)
with tf.Session() as sess:
init_op = tf.global_variables_initializer()
sess.run(init_op)
STEPS = 40000
for i in range(STEPS):
start = (i*BATCH_SIZE) % 300
end = start + BATCH_SIZE
sess.run(train_step, feed_dict={x: X[start:end], y_:Y_[start:end]})
if i % 2000 == 0:
loss_v = sess.run(loss_total, feed_dict={x:X,y_:Y_})
print("After %d steps, loss is: %f" %(i, loss_v))
xx, yy = np.mgrid[-3:3:.01, -3:3:.01]
grid = np.c_[xx.ravel(), yy.ravel()]
probs = sess.run(y, feed_dict={x:grid})
probs = probs.reshape(xx.shape)
print ("w1:\n",sess.run(w1))
print ("b1:\n",sess.run(b1))
print ("w2:\n",sess.run(w2))
print ("b2:\n",sess.run(b2))
plt.scatter(X[:,0], X[:,1], c=np.squeeze(Y_c))
plt.contour(xx, yy, probs, levels=[.5])
plt.show()
![图片.png](https://img.haomeiwen.com/i11956727/42599f6d8e888f78.png?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
网友评论