No gradients provided for any variable - machine-learning

Question

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
Nclass = 500
D = 2
M = 3
K = 3
X1 = np.random.randn(Nclass, D) + np.array([0, -2])
X2 = np.random.randn(Nclass, D) + np.array([2, 2])
X3 = np.random.randn(Nclass, D) + np.array([-2, 2])
X = np.vstack ([X1, X2, X3]).astype(np.float32)
Y = np.array([0]*Nclass + [1]*Nclass + [2]*Nclass)
plt.scatter(X[:,0], X[:,1], c=Y, s=100, alpha=0.5)
plt.show()
N = len(Y)
T = np.zeros((N, K))
for i in range(N):
T[i, Y[i]] = 1
def init_weights(shape):
return tf.Variable(tf.random_normal(shape, stddev=0.01))
def forward(X, W1, b1, W2, b2):
Z = tf.nn.sigmoid(tf.matmul(X, W1) + b1)
return tf.matmul(Z, W2) + b2
tfX = tf.placeholder(tf.float32, [None, D])
tfY = tf.placeholder(tf.float32, [None, K])
W1 = init_weights([D, M])
b1 = init_weights([M])
W2 = init_weights([M, K])
b2 = init_weights([K])
py_x = forward(tfX, W1, b1, W2, b2)
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=py_x, logits=T))
train_op = tf.train.GradientDescentOptimizer(0.05).minimize(cost)
predict_op = tf.argmax(py_x, 1)
sess = tf.Session()
inti = tf.initizalize_all_variables()
for i in range(1000):
sess.run(train_op, feed_dict={tfX: X, tfY: T})
pred = sess.run(predict_op, feed_dict={tfX: X, tfY: T})
if i % 10 == 0:
print(np.mean(Y == pred))
I have a little issue :
Traceback (most recent call last):
File "test.py", line 45, in <module>
train_op = tf.train.GradientDescentOptimizer(0.05).minimize(cost)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/training/optimizer.py", line 322, in minimize
([str(v) for _, v in grads_and_vars], loss))
ValueError: No gradients provided for any variable, check your graph for ops that do not support gradients, between variables ["<tf.Variable 'Variable:0' shape=(2, 3) dtype=float32_ref>", "<tf.Variable 'Variable_1:0' shape=(3,) dtype=float32_ref>", "<tf.Variable 'Variable_2:0' shape=(3, 3) dtype=float32_ref>", "<tf.Variable 'Variable_3:0' shape=(3,) dtype=float32_ref>"] and loss Tensor("Mean:0", shape=(), dtype=float64).
It is unclear what I have to do here. Could anyone be able to help me at this point?

Answer 1

If T are the true labels and py_x the network outputs, you will have to switch the arguments in the cross entropy function:
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=T, logits=py_x))
The logits must be the network outputs and the labels must be the true labels. If you confuse the arguments, the optimizer will fail to backpropagate, since there will be no gradient.
You also have to initialize your variables before training; your code lacks a sess.run(init) statement (you also had a typo in your initialize_all_variables().
I also shuffled your data; maybe it will lead to faster convergence towards the labels.
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
Nclass = 500
D = 2
M = 3
K = 3
X1 = np.random.randn(Nclass, D) + np.array([0, -2])
X2 = np.random.randn(Nclass, D) + np.array([2, 2])
X3 = np.random.randn(Nclass, D) + np.array([-2, 2])
X = np.vstack ([X1, X2, X3]).astype(np.float32)
Y = np.array([0]*Nclass + [1]*Nclass + [2]*Nclass)
perm = np.random.permutation(len(X))
X = X[perm]
Y = Y[perm]
# plt.scatter(X[:,0], X[:,1], c=Y, s=100, alpha=0.5)
# plt.show()
N = len(Y)
T = np.zeros((N, K))
for i in range(N):
T[i, Y[i]] = 1
print(T)
def init_weights(shape):
return tf.Variable(tf.random_normal(shape, stddev=0.01))
def forward(X, W1, b1, W2, b2):
Z = tf.nn.sigmoid(tf.matmul(X, W1) + b1)
return tf.matmul(Z, W2) + b2
tfX = tf.placeholder(tf.float32, [None, D])
tfY = tf.placeholder(tf.float32, [None, K])
W1 = init_weights([D, M])
b1 = init_weights([M])
W2 = init_weights([M, K])
b2 = init_weights([K])
py_x = forward(tfX, W1, b1, W2, b2)
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=T, logits=py_x))
train_op = tf.train.GradientDescentOptimizer(0.1).minimize(cost)
predict_op = tf.argmax(py_x, 1)
sess = tf.Session()
init = tf.initialize_all_variables()
sess.run(init)
for i in range(1000):
sess.run(train_op, feed_dict={tfX: X, tfY: T})
pred = sess.run(predict_op, feed_dict={tfX: X, tfY: T})
if i % 10 == 0:
print(np.mean(Y == pred))

Answer 2

It figured out that you should run inti i.e.
inti = tf.initialize_all_variables()
sess.run(inti)
before running the GradientDescentOptimizer