For an assignment, I'm supposed to write a single layer neural network for one part of it. I think I got most of the stuff right, however when I tried using the tf.nn.softmax_cross_entropy_with_logits method, I got an error saying "ValueError: Both labels and logits must be provided." Which obviously means I need to provide both labels and logits, as I only provided logits in my code right now, so I understand what is wrong. What I don't understand is, what is labels and how I use them in this context? Keep in mind that I'm fairly new and inexperienced in tensorflow and neural networks in general. Thanks!
In supervised learning you have to give labels along with the training data and softmax_cross_entropy_with_logits calculates the softmax cross entropy between logits and labels. It helps to give the probability of a data being in a particular class. You can read more about it here https://www.tensorflow.org/api_docs/python/tf/nn/softmax_cross_entropy_with_logits
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
I've given you a snippet of code from tensorflow tutorials wheresoftmax_cross_entropy_with_logits is used. Here y_ is a placeholder to which the labels are fed.
Also softmax_cross_entropy_with_logits is currently deprecated.