神经网络优化中,自定义损失函数更灵活
例如酸奶销售,生产多了损失成本,生产少了,损失利润,假设生产成本COST为1元,利润PROFIT为9 元,实际生产数量为y,市场需求量为y_
则损失为分段函数,损失为 loss = ( y - y_ ) *COST ( 条件为:y > y- ),loss = ( y_ - y ) * PROFIT (条件为: y_ > y )
python 里可以通过这行代码来完成这个计算 :
loss = tf.reduce_sum(tf.where(tf.greater(y,y_),(y-y_)*COST,(y_-y)*PROFIT))
下面是全部代码:
#coding:utf-8
import tensorflow as tf
import numpy as np
#损失函数
BATCH_SIZE=8
seed = 23455
COST = 1
PROFIT = 9
#基于seed产生随机数
rng = np.random.RandomState(seed)
#随机数返回32行2列的矩阵,作为输入数据集
X = rng.rand(32,2)
#从X矩阵中取出一行,求和后判断结果,小于1给Y赋值1,不大于1给Y赋值0
#Y作为输入数据集的标签(正确结果)
Y = [[(x0 + x1+rng.rand()/10.0-0.05)] for (x0,x1) in X]
print("X:\n",X)
print("Y:\n",Y)
#定义神经网络的输入、参数和输出,定义向前传播过程
x=tf.placeholder(tf.float32,shape=(None,2))
y_=tf.placeholder(tf.float32,shape=(None,1))
w1 = tf.Variable(tf.random_normal([2,1],stddev=1,seed=1))
y = tf.matmul(x,w1)
#定义损失函数及反向传播方法
#loss = tf.reduce_mean(tf.square(y-y_))
loss = tf.reduce_sum(tf.where(tf.greater(y,y_),(y-y_)*COST,(y_-y)*PROFIT))
train_step = tf.train.GradientDescentOptimizer(0.001).minimize(loss)
#train_step = tf.train.MomentumOptimizer(0.001,0.9).minimize(loss)
#train_step = tf.train.AdamOptimizer(0.001).minimize(loss)
#生成会话,训练STEPS轮
with tf.Session() as sess :
init_op = tf.global_variables_initializer()
sess.run(init_op)
#训练模型
STEPS = 60000
for i in range(STEPS):
start = (i*BATCH_SIZE)%32
end = start + BATCH_SIZE
sess.run(train_step,feed_dict={x:X[start:end],y_:Y[start:end]})
if i % 500 ==0:
print("After %d steps the w1 is: "%i)
print(sess.run(w1),"\n")
print("the last result w1 is :")
print(sess.run(w1),"\n")
运行截图
这是我的github源码链接:https://github.com/jiang-congcong/Deap-learning-simple-NN_2