Runs CTC loss algorithm on each batch element.

`k_ctc_batch_cost(y_true, y_pred, input_length, label_length)`

Tensor with shape (samples,1) containing the CTC loss of each element.

- y_true
tensor

`(samples, max_string_length)`

containing the truth labels.- y_pred
tensor

`(samples, time_steps, num_categories)`

containing the prediction, or output of the softmax.- input_length
tensor

`(samples, 1)`

containing the sequence length for each batch item in`y_pred`

.- label_length
tensor

`(samples, 1)`

containing the sequence length for each batch item in`y_true`

.

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://tensorflow.rstudio.com/reference/keras/index.html#backend.