Upstream missing jaxopt losses to optax - Part 2/N

Adding a deprecated alias for the names that `softmax_cross_entropy_with_integer_labels` and `sigmoid_binary_cross_entropy` have in `jaxopt` to make it easier for users moving over to find the corresponding loss in optax.

PiperOrigin-RevId: 617196448

optax/losses/_classification.py

@@ -60,6 +60,13 @@ class is an independent binary prediction and different classes are not
   return -labels * log_p - (1. - labels) * log_not_p
 
 
+@functools.partial(
+    chex.warn_deprecated_function,
+    replacement='sigmoid_binary_cross_entropy')
+def binary_logistic_loss(logits, labels):
+  return sigmoid_binary_cross_entropy(logits, labels)
+
+
 def hinge_loss(
     predictor_outputs: chex.Array,
     targets: chex.Array
@@ -76,6 +83,25 @@ def hinge_loss(
   return jnp.maximum(0, 1 - predictor_outputs * targets)
 
 
+def perceptron_loss(
+    predictor_outputs: chex.Numeric,
+    targets: chex.Numeric
+) -> chex.Numeric:
+  """Binary perceptron loss.
+
+  References:
+    https://en.wikipedia.org/wiki/Perceptron
+
+  Args:
+    predictor_outputs: score produced by the model (float).
+    targets: Target values. Target values should be strictly in the set {-1, 1}.
+
+  Returns:
+    loss value.
+  """
+  return jnp.maximum(0, - predictor_outputs * targets)
+
+
 def softmax_cross_entropy(
     logits: chex.Array,
     labels: chex.Array,
@@ -139,6 +165,13 @@ def softmax_cross_entropy_with_integer_labels(
   return log_normalizers - label_logits
 
 
+@functools.partial(
+    chex.warn_deprecated_function,
+    replacement='softmax_cross_entropy_with_integer_labels')
+def multiclass_logistic_loss(logits, labels):
+  return softmax_cross_entropy_with_integer_labels(logits, labels)
+
+
 @functools.partial(chex.warn_only_n_pos_args_in_future, n=2)
 def poly_loss_cross_entropy(
     logits: chex.Array,