Runs CTC loss algorithm on each batch element.
k_ctc_batch_cost(y_true, y_pred, input_length, label_length)tensor (samples, max_string_length) containing the truth
labels.
tensor (samples, time_steps, num_categories) containing the
prediction, or output of the softmax.
tensor (samples, 1) containing the sequence length for
each batch item in y_pred.
tensor (samples, 1) containing the sequence length for
each batch item in y_true.
Tensor with shape (samples,1) containing the CTC loss of each element.
This function is part of a set of Keras backend functions that enable lower level access to the core operations of the backend tensor engine (e.g. TensorFlow, CNTK, Theano, etc.).
You can see a list of all available backend functions here: https://keras.rstudio.com/articles/backend.html#backend-functions.