Computes the crossentropy metric between the labels and predictions

metric_categorical_crossentropy(
  y_true,
  y_pred,
  from_logits = FALSE,
  label_smoothing = 0,
  axis = -1L,
  ...,
  name = "categorical_crossentropy",
  dtype = NULL
)

Arguments

y_true

Tensor of true targets.

y_pred

Tensor of predicted targets.

from_logits

(Optional) Whether output is expected to be a logits tensor. By default, we consider that output encodes a probability distribution.

label_smoothing

(Optional) Float in [0, 1]. When > 0, label values are smoothed, meaning the confidence on label values are relaxed. e.g. label_smoothing=0.2 means that we will use a value of 0.1 for label 0 and 0.9 for label 1"

axis

(Optional) (1-based) Defaults to -1. The dimension along which the metric is computed.

...

Passed on to the underlying metric. Used for forwards and backwards compatibility.

name

(Optional) string name of the metric instance.

dtype

(Optional) data type of the metric result.

Value

If y_true and y_pred are missing, a (subclassed) Metric instance is returned. The Metric object can be passed directly to compile(metrics = ) or used as a standalone object. See ?Metric for example usage.

Alternatively, if called with y_true and y_pred arguments, then the computed case-wise values for the mini-batch are returned directly.

Details

This is the crossentropy metric class to be used when there are multiple label classes (2 or more). Here we assume that labels are given as a one_hot representation. eg., When labels values are c(2, 0, 1):

 y_true = rbind(c(0, 0, 1),
                c(1, 0, 0),
                c(0, 1, 0))`

See also