我正在训练一个判别网络，以便在生成网络中使用。我使用包含两个特征的数据集进行训练，并进行二元分类。1 表示冥想，0 表示未冥想。（数据集来自@[隐藏人名]的一个视频）。
由于某些原因，输出层（ol）在每个测试用例中总是输出 [1]。
我的数据集： https://drive.google.com/open?id=0B5DaSp-aTU-KSmZtVmFoc0hRa3c
import pandas as pdimport tensorflow as tfdata = pd.read_csv("E:/workspace_py/datasets/simdata/linear_data_train.csv")data_f = data.drop("lbl", axis = 1)data_l = data.drop(["f1", "f2"], axis = 1)learning_rate = 0.01batch_size = 1n_epochs = 30n_examples = 999 # This is highly unsatisfying >:3n_iteration = int(n_examples/batch_size)features = tf.placeholder('float', [None, 2], name='features_placeholder')labels = tf.placeholder('float', [None, 1], name = 'labels_placeholder')weights = { 'ol': tf.Variable(tf.random_normal([2, 1], stddev= -12), name = 'w_ol')}biases = { 'ol': tf.Variable(tf.random_normal([1], stddev=-12), name = 'b_ol')}ol = tf.nn.sigmoid(tf.add(tf.matmul(features, weights['ol']), biases['ol']), name = 'ol')loss = -tf.reduce_sum(labels*tf.log(ol), name = 'loss') # cross entropytrain = tf.train.AdamOptimizer(learning_rate).minimize(loss)sess = tf.Session()sess.run(tf.global_variables_initializer())for epoch in range(n_epochs): ptr = 0 for iteration in range(n_iteration): epoch_x = data_f[ptr: ptr + batch_size] epoch_y = data_l[ptr: ptr + batch_size] ptr = ptr + batch_size _, err = sess.run([train, loss], feed_dict={features: epoch_x, labels:epoch_y}) print("Loss @ epoch ", epoch, " = ", err)print("Testing...\n")data = pd.read_csv("E:/workspace_py/datasets/simdata/linear_data_eval.csv")test_data_l = data.drop(["f1", "f2"], axis = 1)test_data_f = data.drop("lbl", axis = 1)#vvvHERE print(sess.run(ol, feed_dict={features: test_data_f})) #<<<HERE#^^^HEREsaver = tf.train.Saver()saver.save(sess, save_path="E:/workspace_py/saved_models/meditation_disciminative_model.ckpt")sess.close()输出:
2017-10-11 00:49:47.453721: W C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.2017-10-11 00:49:47.454212: W C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\platform\cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.2017-10-11 00:49:49.608862: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:955] Found device 0 with properties: name: GeForce GTX 960Mmajor: 5 minor: 0 memoryClockRate (GHz) 1.176pciBusID 0000:01:00.0Total memory: 4.00GiBFree memory: 3.35GiB2017-10-11 00:49:49.609281: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:976] DMA: 0 2017-10-11 00:49:49.609464: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:986] 0: Y 2017-10-11 00:49:49.609659: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\35\tensorflow\core\common_runtime\gpu\gpu_device.cc:1045] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 960M, pci bus id: 0000:01:00.0)Loss @ epoch 0 = 0.000135789Loss @ epoch 1 = 4.16049e-05Loss @ epoch 2 = 1.84776e-05Loss @ epoch 3 = 9.41758e-06Loss @ epoch 4 = 5.24522e-06Loss @ epoch 5 = 2.98024e-06Loss @ epoch 6 = 1.66893e-06Loss @ epoch 7 = 1.07288e-06Loss @ epoch 8 = 5.96047e-07Loss @ epoch 9 = 3.57628e-07Loss @ epoch 10 = 2.38419e-07Loss @ epoch 11 = 1.19209e-07Loss @ epoch 12 = 1.19209e-07Loss @ epoch 13 = 1.19209e-07Loss @ epoch 14 = -0.0Loss @ epoch 15 = -0.0Loss @ epoch 16 = -0.0Loss @ epoch 17 = -0.0Loss @ epoch 18 = -0.0Loss @ epoch 19 = -0.0Loss @ epoch 20 = -0.0Loss @ epoch 21 = -0.0Loss @ epoch 22 = -0.0Loss @ epoch 23 = -0.0Loss @ epoch 24 = -0.0Loss @ epoch 25 = -0.0Loss @ epoch 26 = -0.0Loss @ epoch 27 = -0.0Loss @ epoch 28 = -0.0Loss @ epoch 29 = -0.0Testing...[[ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.] [ 1.]]Saving model...[Finished in 57.9s]回答：
主要问题
首先，这不是有效的交叉熵损失。你使用的方程仅在有两个或更多输出时有效。对于单个sigmoid输出，你必须这样做
-tf.reduce_sum(labels*tf.log(ol) + (1-labels)*tf.log(1-ol), name = 'loss')否则，最优解总是回答“1”（这正是现在发生的情况）。
为什么？
请注意，标签只有0或1，你的整个损失是标签与预测的对数的乘积。因此，当真实标签为0时，无论你的预测是什么，损失都是0，因为0 * log(x) = 0，无论x是什么（只要log(x)定义了）。因此，你的模型只会在应该预测“1”时未预测“1”时受到惩罚，所以它学会了一直输出1。
其他一些奇怪之处
-
你为正态分布提供了负的stddev，而你不应该这样做（除非这是random_normal的一些未记录的功能，但根据文档，它应该接受一个单一的正浮点数，你应该在那里提供一个小的数字）。
-
像这样计算交叉熵（以一种天真的方式）在数值上不稳定，请查看tf.sigmoid_cross_entropy_with_logits。
-
你没有对数据集进行排列，因此你总是以相同的顺序处理数据，这可能会产生不良后果（损失的周期性增加，收敛困难或无法收敛）。
