我正在学习 tf.contrib.opt.ScipyOptimizerInterface 并正在开发一个演示,其中每个权重都是非负的。
# -*- coding: utf-8 -*-
import tensorflow as tf
vector = tf.constant([.021,.046,.013], name='vector')
wt = tf.Variable([1./3,1./3,1./3], 'wt')
loss = -tf.reduce_sum(tf.multiply(vector,wt,'loss'))
equalities = [tf.reduce_sum(wt) - 1.]
inequalities = [wt[0],wt[1],wt[2]]
optimizer = tf.contrib.opt.ScipyOptimizerInterface(loss, var_list=[wt], equalities=equalities, inequalities=inequalities, method='SLSQP')
with tf.Session() as session:
session.run(tf.global_variables_initializer())
optimizer.minimize(session)
equalities: Optional list of equality constraint scalar Tensors to be held equal to zero.inequalities: Optional list of inequality constraint scalar Tensors to be kept nonnegative.
如何将 inequalities = [wt[0],wt[1],wt[2]] 更改为 inequalities = [wt[i] for i in range( tf.size(weight))]?
请您参考如下方法:
你可以设置它:
inequalities = [wt[i] for i in range(wt.get_shape()[0])]
