Determine pinball loss function in kera using backend
I am trying to define a pinbal loss function to implement "quantile regression" in a neural network with Keras (with Tensorflow as the backend).
The definition is here: pinball loss
It is difficult to implement the traditional K.means () function, etc. since they deal with the whole batch of y_pred, y_true, but I have to consider each component y_pred, y_true and here's my source code:
def pinball_1(y_true, y_pred):
loss = 0.1
with tf.Session() as sess:
y_true = sess.run(y_true)
y_pred = sess.run(y_pred)
y_pin = np.zeros((len(y_true), 1))
y_pin = tf.placeholder(tf.float32, [None, 1])
for i in range((len(y_true))):
if y_true[i] >= y_pred[i]:
y_pin[i] = loss * (y_true[i] - y_pred[i])
else:
y_pin[i] = (1 - loss) * (y_pred[i] - y_true[i])
pinball = tf.reduce_mean(y_pin, axis=-1)
return K.mean(pinball, axis=-1)
sgd = SGD(lr=0.1, clipvalue=0.5)
model.compile(loss=pinball_1, optimizer=sgd)
model.fit(Train_X, Train_Y, nb_epoch=10, batch_size=20, verbose=2)
I tried to pass y_pred, y_true into a vector data structure, so I can cast them with an index and handle the individual components, but it seems the problem comes from a lack of knowledge about handling y_pred, y_true individually.
I tried to dive into error related lines, but I was almost lost.
InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'dense_16_target' with dtype float
[[Node: dense_16_target = Placeholder[dtype=DT_FLOAT, shape=[], _device="/job:localhost/replica:0/task:0/cpu:0"]()]]
How can I fix this? Thank!
+3
source to share