我已经定义了一个TensorFlow神经网络,如下所示:
W1 = tf.Variable(tf.truncated_normal([no_features, neurons_hd1], stddev=0.03), name='W1')
b1 = tf.Variable(tf.truncated_normal([neurons_hd1]), name='b1')
...
W5 = tf.Variable(tf.truncated_normal([neurons_hd4, no_outputs], stddev=0.03), name='W5')
b5 = tf.Variable(tf.truncated_normal([no_outputs]), name='b4')
hidden1_out = tf.nn.relu(tf.add(tf.matmul(x, W1), b1))
...
y_ = tf.nn.tanh(tf.add(tf.matmul(hidden4_out, W5), b5))
loss = tf.losses.mean_squared_error(y, y_)
optimizer = tf.train.GradientDescentOptimizer(learning_rate=learning_rate).minimize(loss)
然后我使用以下方法进行训练:
... = sess.run([optimizer, loss], feed_dict={x: batch_x, y: batch_y})
我想申请“手动”退学;我需要从W1,W2,...,W5务实地挑选“特定”权重以退出。如何使用TensorFlow做到这一点?